Kate Strachnyi has a habit of calling people out on LinkedIn. When she spots a post that’s been run through an AI rewriter, she’ll send the person a direct message. “Hey, I could tell you used AI,” she’ll say. The usual response is some version of “but these are my thoughts,” and her advice back is to keep them that way and stop running them through the machine. She calls that keeping your content “non-GMO,” unmodified, and authentic.

It’s a funny line, and it also describes her entire business model. Kate is the founder of DATAcated, a media company that partners with brands in data, analytics, and AI to reach their audiences through real content creators and thought leaders. In a market where AI-generated posts are flooding every feed, and ironically, LinkedIn itself has a “rewrite with AI” button baked into the platform, Kate is making the opposite bet. She’s building a business around real people with real expertise and real opinions.

On Episode 38 of the Data Faces Podcast, I sat down with Kate to talk about how she built DATAcated from a one-person experiment into an influencer agency with 40+ creators, why she’s doubling down on authenticity as AI content takes over, and what happens to expertise itself when the humans who hold it stop doing the work.

“Keep it non-GMO. Don’t modify your content, just leave it as is.”

— Kate Strachnyi, Founder, DATAcated

About Kate Strachnyi

Kate Strachnyi is the founder of DATAcated. She started her career in finance and risk management consulting before pivoting to data visualization 12 years ago. Kate has since written five books, established one of the most connected networks of data and AI professionals in the industry, and built DATAcated into a full agency, with 40+ influencers, speakers, and subject-matter experts through her DATAcated Plus program. She is a LinkedIn Top Voice. In our conversation on Episode 38 of the Data Faces Podcast, we discuss:

- How Kate followed the revenue data from courses and books to a focused media business

- The DATAcated Plus model and how influencer campaigns work behind the scenes

- Why she’s shifting from “Kate = DATAcated” to an agency brand

- The flood of AI-generated content on LinkedIn and her “non-GMO” content philosophy

- What happens to expertise when today’s subject matter experts retire and AI fills the void

Following the data from finance to media

Kate’s path to running a media company started with a practical constraint. She was working in financial services risk management at a large consulting firm, traveling Monday through Thursday, and expecting her first child. She spent eight months searching for any role that would let her work remotely, well before remote work was mainstream. Someone eventually pointed her toward a data role that involved Tableau, visualization, and “creating pretty pictures.” She took it and fell in love with data storytelling.

“I am a data person, right? So I would look at the numbers and see what is driving more revenue and what is allowing me to have more time to myself.”

— Kate Strachnyi, Founder, DATAcated

What followed was a period of deliberate experimentation. Kate wrote books and launched an academy. She created courses, ran her own conferences, and built a community called DATAcated Circle. As a business of one with nobody to stop her, she could try anything, and she tracked the results. The revenue data and the work she enjoyed most both pointed to media and content creation. That’s where DATAcated lives today.

Building DATAcated Plus — from personal brand to influencer agency

When Kate started getting more client work than she could handle, she brought in the content creators she already knew. DATAcated Plus was born.

The program now spans data, analytics, and AI, with the agentic AI space growing fastest. When a client comes to Kate with a product launch, an event, or a brand awareness campaign, she matches them with creators based on their audience, expertise, and the success metrics the client is targeting. She reviews every piece of content before it goes to the client for approval. The day-to-day project management runs underneath, with UTM links, calls to action, and timelines coordinated across every moving piece. None of the work is glamorous, but it’s what keeps the whole operation running.

The DATAcated Plus roster also includes speakers and experts that companies can hire for keynotes and panels, as well as webinars and thought leadership papers. Kate has placed people across major industry events, including multiple years at the Gartner Data & Analytics Summit, where her team creates what she calls “FOMO-inducing content” on-site.

Scott Taylor, the Data Whisperer, joined Kate at last year’s Gartner event to co-lead a sold-out breakfast session on personal branding for data leaders. One question from the audience: “I want to post, but my company said no.” Kate’s advice was to follow your company’s rules but find the leeway. Talk about your perspective on industry topics rather than your specific projects or tools. When there’s no leeway at all, I suggested it might be time to find a company that sees your personal brand as an asset rather than a risk.

At Big Data London, a group of DATAcated Plus creators went to a tattoo parlor and got fake data tattoos for a video so convincing that Kate’s neighbors congratulated her on the new ink.

“People will unfollow instantly if we just keep promoting things. It’s more of letting my audience know about here’s a product that exists, and here’s what it does, in case you need it.”

— Kate Strachnyi, Founder, DATAcated

That creative range is what separates the DATAcated model from a traditional analyst engagement. Analyst firms produce authoritative research with independent evaluations, and content creators bring flexibility, personality, and a wider range of formats from short-form video to live event coverage.^[1] Kate’s crew has done cooking shows to explain data governance and built sandcastles to illustrate strong data foundations. The content sticks with audiences, and brands gain reach from creators who have built genuine trust over years of showing up with their own voices.

More recently, Kate has changed how she positions the company itself. The old DATAcated media kit led with a big photo of Kate and a rundown of what she could do. The new version leads with the influencer roster. She’s deliberately moving from “Kate equals DATAcated” to an agency brand that doesn’t depend on her being in every room. Some clients still ask for Kate specifically, and she’s learning to redirect them toward creators who are a better fit. “It’s nice to be wanted,” she told me, “but it doesn’t scale.”

The authenticity bet in an AI-saturated feed

Kate is leaning harder into genuine human voices at the exact moment AI-generated content is flooding LinkedIn and every other platform. LinkedIn added a “rewrite with AI” button to the post editor, while its own users complain that AI-generated slop is taking over their feeds. Kate’s view is that if you know a person well enough, you can spot AI-written content immediately. The vocabulary is off, the phrasing is too polished, and the voice sounds like everyone else’s. She calls people out on it, and she expects the same standard from her DATAcated Plus creators. For Kate, authentic content means it was written by the credited human, reflects their expertise and opinions, and hasn’t been reprocessed through an AI rewriter.

That doesn’t mean she’s anti-AI. Kate recently spent an entire day automating her invoicing process in Claude Code. The task itself takes two minutes, but she never has to do it by hand again. She and I compared notes about automating YouTube uploads, scheduling content, and eliminating the repetitive copy-paste work that eats up a solopreneur’s day. She draws the line between back-office operations and audience-facing content. AI can handle the invoices. It should not rewrite your LinkedIn posts.

“What are we going to do 20 years from now, when we don’t have those subject matter experts? They’re retired or not working anymore. Because if you don’t work with this stuff, whatever that might be in the medical field, in construction, how are you going to fact-check it?”

— Kate Strachnyi, Founder, DATAcated

Part of our discussion focused on a question Kate had heard at a recent event. Right now, subject matter experts can look at AI-generated output and spot what’s wrong because they’ve spent decades doing the work firsthand. But what happens in 20 years, when those experts have retired? If the next generation learns from AI output instead of from direct experience, the ability to verify and correct that output disappears.

The humans who make genuine content valuable are also the humans who keep AI honest. Kate sees her business as part of the answer. Invest in real experts now, amplify their voices, and make sure the knowledge doesn’t evaporate into a feedback loop of machine-generated content. The window to establish yourself as a real authority, someone whose voice carries weight because it’s grounded in lived experience, won’t stay open forever.

I started the Data Faces Podcast to have real conversations with the people doing the work. The messy, honest, sometimes funny exchanges that you can only get from humans who have opinions and aren’t afraid to share them. Kate’s business is built on the same conviction. In a world filling up with synthetic content, she’s betting that real voices will only become more valuable. When I asked her about deepfakes and AI versions of herself, her answer was four words.

“I plan to remain authentic.”

— Kate Strachnyi, Founder, DATAcated

Listen to the full conversation with Kate Strachnyi on the Data Faces Podcast.

Based on insights from Kate Strachnyi, Founder at DATAcated, featured on the Data Faces Podcast.

Leave a comment

Frequently asked questions

What is DATAcated, and what does the company do?

DATAcated is a media company founded by Kate Strachnyi, author of five books, including ColorWise and Journey to Data Scientist, that helps brands in data, analytics, and AI reach their target audiences through authentic content creators and thought leaders. The company runs the DATAcated Plus program, a roster of 40+ influencers, speakers, and subject-matter experts who create thought-leadership content, amplify product launches, and represent brands at industry events such as the Gartner Data & Analytics Summit. DATAcated operates as an agency that matches creators to client campaigns based on audience fit and success metrics.

What is the DATAcated Plus program?

DATAcated Plus is Kate Strachnyi’s influencer and speaker program for the data and AI industry. It includes content creators, speakers, and subject matter experts across data analytics, AI, and agentic AI. Companies hire DATAcated Plus members for brand awareness campaigns, event coverage, webinars, thought leadership papers, and on-site content creation. Kate manages the program by vetting creators for authenticity, reviewing all content before client approval, and coordinating timelines and deliverables across campaigns.

How does influencer marketing differ from analyst relations in B2B tech?

Analyst firms like Gartner and Forrester produce authoritative research and independent evaluations. Influencer content creators offer more creative flexibility and can be guided toward specific messaging for a campaign. Kate Strachnyi’s DATAcated Plus creators have done cooking shows to explain data governance and built sandcastles to illustrate data foundations. They’ve also produced viral video content at industry events. Both approaches build credibility with B2B audiences, and many companies now use influencers and analysts together at the same events.

What does “non-GMO content” mean in the context of AI and social media?

“Non-GMO content” is Kate Strachnyi’s phrase for content that hasn’t been run through an AI rewriter. Just as non-GMO food is unmodified, non-GMO content preserves the author’s original voice and phrasing. Kate advocates for this approach because AI-rewritten posts lose the personality and authenticity that make content creators valuable to their audiences. She actively calls out AI-washed posts on LinkedIn and holds her DATAcated Plus creators to the same standard.

Will AI replace subject matter experts in data and AI content?

Kate Strachnyi raises a concern about long-term expertise. Today’s subject matter experts can spot errors in AI-generated content because they have decades of hands-on experience. In 20 years, when those experts have retired, the ability to fact-check and verify AI output may disappear if the next generation learns from AI-generated content rather than direct experience. Kate’s business model is built around investing in real human experts now and amplifying their voices before that institutional knowledge erodes.

Podcast highlights

[0:05] Kate’s background and what DATAcated does

[2:10] Pre-finance Kate: what she wanted to be before data found her

[3:05] The career pivot from risk management consulting to data visualization

[5:03] How DATAcated evolved from training and books to a focused media company

[7:27] How the influencer model works behind the scenes

[9:33] Automating business operations with Claude Code

[11:01] Walking the line between brand amplification and spam

[14:11] The fake tattoo story from Big Data London

[15:03] How DATAcated Plus compares to analyst firm engagements

[17:14] The sold-out personal branding session at Gartner with Scott Taylor

[22:15] Shifting from “Kate = DATAcated” to an agency brand

[24:02] What works on LinkedIn now vs. five years ago

[27:01] AI-generated content flooding feeds and the “non-GMO” philosophy

[29:04] The 20-year question: who fact-checks AI when the experts retire?

[30:20] Deep fake Dave and why Kate plans to remain authentic

[31:24] Why Kate hasn’t hired a team and is betting on AI for operations

[33:57] Does AI make you more productive or just busier?

[36:19] Where to find Kate and DATAcated

About David Sweenor

David Sweenor is the founder and host of the Data Faces podcast, where he talks with the people who are making data, analytics, AI, and marketing work in the real world. He is also the founder of TinyTechGuides and a recognized top 25 analytics thought leader and international speaker who specializes in practical business applications of artificial intelligence and advanced analytics.

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence. David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Books

- Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

- Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

- The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

- The CIO’s Guide to Adopting Generative AI: Five Keys to Success

- Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

- The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn.

^[1]Sweenor, David. “Stop Writing AI Content That Sounds Like Everyone Else’s.” TinyTechGuides, February 7, 2025. https://insights.tinytechguides.com/p/stop-writing-ai-content-that-sounds

For 25 years, data quality has been everyone’s problem and nobody’s priority. For some, it was an IT problem, and for others, it was a business problem. But most of the time, fixing it at scale was largely ignored. What would you do if a number in the spreadsheet looked off? You’d fix it and move on with your day. The same with questionable metrics on dashboards. We’ve been able to tuck and hide the cost of bad data in a manual world for a while now. Since the pace of business was slower, there were no real consequences for getting it wrong.

Those ways of old change when you hand autonomy to an AI agent. An agent doesn’t pause to gut-check a suspicious number, it doesn’t really care. It takes the data at face value, makes a decision, feeds that decision into the next step, and keeps going. You might be six or seven steps down the line before anyone realizes the foundation was wrong. And by then, the damage compounds in ways that a quick spreadsheet fix can’t undo.

I sat down with Brendan Grady, EVP and General Manager of Analytics and AI at Qlik, at Qlik Connect 2026 in Orlando to discuss why the stakes around data quality have changed, where enterprise-agentic adoption stands today, and what data professionals should be thinking about.

“In today’s world where there may be an agent running around using said data and getting it wrong, the consequences of getting it wrong are going to be catastrophic.”

— Brendan Grady, EVP and GM of Analytics & AI, Qlik

About Brendan Grady

Brendan Grady is EVP and General Manager of the Analytics and AI Business Unit at Qlik, where he leads product management, product design, R&D, and go-to-market strategy for the company’s data integration, quality, and analytics platform. Before Qlik, he held senior GTM roles at IBM, where he led worldwide digital sales for Watson Analytics and managed the Cognos portfolio. He joined Qlik seven years ago after repeatedly losing deals to its analytics engine, and decided to find out why. And well before all of that, he delivered the Sound of Music tour in Salzburg, Austria, over 300 times.

In this episode, we discuss:

- Why data quality was never fixed and why that matters now

- Where enterprise agentic AI adoption actually stands

- Trust scores and the problem with feeding spreadsheets to LLMs

- The shift from dashboards to decision intelligence

- Open standards, MCP, and why there’s no “One Ring to rule them all”

- Advice for data professionals navigating the AI transition

The consequence management problem

Grady framed the data quality conversation in a way I hadn’t heard before. He called it consequence management. For decades, organizations tolerated bad data because the consequences of getting it wrong were manageable. A field was incorrect in a report? Someone caught it, fixed it, and everyone moved on, knowing there would be another fire drill tomorrow. The recovery cost was low enough that nobody prioritized prevention, and if they did, they rarely had the organizational backing to make any meaningful change.

“Is it IT’s job? Is it the business’s job? Is it both, or is it nobody’s job? For most companies, it’s been nobody’s job.”

— Brendan Grady, EVP and GM of Analytics & AI, Qlik

BARC’s research confirms this pattern. As Shawn Rogers discussed on the Data Faces Podcast, data quality remains the top challenge for organizations trying to mature their analytics and AI capabilities.^[1] That organizational ambiguity persisted because the stakes allowed it. He pointed to real examples. A major airline took a significant hit to its market cap because its sentiment data was wrong and decisions were made on flawed analysis. Two decades ago, a single field in a spreadsheet contributed to a financial crisis that rippled through an entire market. These weren’t hypothetical scenarios. They happened because nobody owned the problem and the systems in place couldn’t detect the errors before they cascaded.

In the agentic era, the failure mode is different. A human looking at a dashboard might notice something feels off and investigate. An agent won’t. It will take the data, reason through it, make a decision, and pass that decision to the next agent in the chain; often without any confidence bounds or trust scores.

The point isn’t that agents are dangerous. The point is that autonomous systems need trusted data underneath them before they’re given the authority to act. Without that bedrock, every step an agent takes amplifies whatever error was baked into the starting point. As practitioners, we know this, why hasn’t this been fixed yet?

“Prior to stage zero”

I asked Brendan where enterprise agentic adoption actually stands. His answer was honest. “What’s prior to stage zero?” he said. “I mean, there are customers that are trying things out there, surely. But from a large-scale production standpoint, we’re in the early days.”

Customers are experimenting with low-risk use cases. They’re testing agents in controlled environments where a mistake won’t damage the business. But production-grade agents making real decisions in real business processes? That’s rare. And the blocker, according to Brendan, isn’t the technology. It’s the data.

Gartner projects that by 2027, 70% of organizations will adopt modern data quality solutions to support AI adoption and digital business initiatives.^[2] That projection tells you where the market is today. If 70% will need to adopt these solutions by 2027, most organizations don’t have them yet. The ambition around agentic AI is running well ahead of the data infrastructure required to support it. Shane Murray made a similar argument on the Data Faces Podcast earlier this year, noting that actionable data strategies beat endless planning when it comes to AI readiness.^[3]

Brendan also raised a practical question that every data leader should be asking. The LLM landscape is shifting constantly. Six months ago it was OpenAI. Today, Claude is gaining traction. Tomorrow the market may have moved on to something new. His advice was to work with vendors that approach this from an open standards perspective, supporting multiple LLMs rather than forcing a single choice. The technology will keep changing, but the data underneath it is what has to hold steady.

“The internet took 10 years, 20 years, 30 years to get going. We’re a year and a half in.”

— Brendan Grady, EVP and GM of Analytics & AI, Qlik

Trust as the missing layer

One of the more revealing moments in our conversation came when Brendan talked about what happens when you feed structured data into an LLM. I’ve experienced this myself. You upload a spreadsheet, ask it to calculate something, and the answer comes back looking polished and confident. The formatting is clean, the language is professional, and unbeknownst to you, the numbers are wrong.

“It’s really pretty, right? The answer is amazing. Looks great. Totally BS. And the next thing you know, you’re showing up to the board with all incorrect numbers.”

— Brendan Grady, EVP and GM of Analytics & AI, Qlik

Qlik’s Trust Score for AI is designed to give decision-makers a quantifiable measure of whether their data is valid, fresh, and representative before it reaches an agent or an LLM.^[4] Instead of hoping your data is accurate, you can see a score that tells you it’s 90% trustworthy or 80% or something that should give you pause.

The other piece Brendan emphasized was intent detection. When someone asks a question of an LLM, the literal question and the actual intent are often different things. I ran into this recently when I asked an AI assistant to analyze several websites. It came back with a confident analysis, but when I pressed it, it admitted it had never actually visited the sites. Qlik is investing in understanding what the user is really trying to accomplish so the system can route to the right data and the right engine rather than letting an LLM fabricate its way to an answer.

The combination of trust scores and intent detection reflects a broader principle. Before you give an agent the authority to act on data, you need to know that the data is sound and that the system understands what you’re actually asking. Qlik’s track record in this space is long. The company has been named a Leader in the Gartner Magic Quadrant for Augmented Data Quality Solutions for seven consecutive years, most recently in 2026.^[5]

“Dashboards are dead. Long live dashboards.”

When Brendan declared that dashboards are dead, I thought I had a scoop and I made sure the audience heard it. He laughed and then walked it back with the nuance that matters. Dashboards as a destination are going away, but the data inside them and the decisions they inform are more important than ever.

Brendan described how his own workflow has changed. He used to ask his analytics tools for information about business performance. Now he asks a different question. “Tell me about my business and what you think I should do.” That shift from information retrieval to decision recommendation is what Qlik means by decision intelligence, and it’s powered by two things working together.

The first is Qlik’s analytics engine, which finds associations and relationships in data that other approaches miss. Instead of running a predefined query to answer a specific question, the engine surfaces connections you didn’t know existed. Brendan called these the unknown unknowns. In an agentic context, that capability becomes even more valuable because it allows agents to explore paths and relationships that a standard SQL query would never surface.

“In the agentic world, we’re serving this up to help agents understand that there’s a relationship here that you need to go explore before you take action. That is extremely powerful.”

— Brendan Grady, EVP and GM of Analytics & AI, Qlik

The second is openness. Qlik launched its MCP server in February 2026, implementing the open Model Context Protocol standard to let third-party AI assistants access Qlik’s analytical capabilities with governance built in.^[6] “There’s never going to be One Ring to rule them all,” he said. People want to work in the tools they’re comfortable with, whether that’s Claude, Gemini, ChatGPT, or something that doesn’t exist yet. The bet is paying off. Brendan shared that they’re already seeing roughly a 50/50 split between users accessing agentic capabilities through Qlik’s own interface and those coming in through MCP.

“Am I out of a job?”

Brendan closed our conversation with a story that we’ve all encountered. After demoing the ability to build an analytics application through Claude in 30 seconds at Qlik Connect, a customer approached him. This person had built his entire career writing code to create analytics applications across multiple platforms. His question was simple. “Am I out of a job?”

Brendan’s answer was no, but with an important caveat. The job will evolve. His advice to data professionals was to lean into what they already know better than anyone else: the data itself. Become the data product owner. Be the trusted guide as organizations navigate the agentic experience. The people who understand the data well enough to know its quirks and business context will be indispensable as agents take on more routine work.

This tracks with what Brendan’s team has seen internally. Qlik has developers who were already performing well, and AI tools have turned them into 10x contributors. The acceleration is happening at the top end, where strong performers are getting faster and producing better work. A preliminary MIT Media Lab study found that heavy reliance on AI assistants can lead to what researchers called “cognitive debt,” where users outsource critical thinking and lose the ability to recall and synthesize what they’ve produced.^[7] Brendan acknowledged this risk directly. He sees his own daughters, 19 and 24, defaulting to LLMs for answers, and he worries about critical thought eroding over time.

“Embrace these new technologies. It’s scary. But your job will evolve. Become that data product owner, become an expert in that data, and be that trusted guide as everybody’s going down the agentic experience.”

— Brendan Grady, EVP and GM of Analytics & AI, Qlik

The real opportunity for data professionals is to become the people who make sure agents are working with the right information in the right context. That’s a role no LLM can fill on its own. If you’re not sure where to start, audit the data your team’s AI tools depend on. If you can’t quantify how trustworthy that data is, that’s the first problem to solve.

Listen to the full conversation with Brendan Grady on the Data Faces Podcast.

Based on insights from Brendan Grady, EVP and GM of Analytics & AI at Qlik, featured on the Data Faces Podcast.

Share

Frequently asked questions

What is consequence management in the context of data quality?

Consequence management is the idea that data quality was never prioritized because the consequences of bad data were manageable. In a manual world, a wrong number in a spreadsheet could be corrected before it caused real damage. With AI agents making autonomous decisions across multiple steps, errors compound before anyone detects them. Consequence management explains why the stakes around data quality have shifted from recoverable inconvenience to potential business-level damage.

Where does enterprise adoption of agentic AI stand in 2026?

According to Brendan Grady, EVP of Analytics and AI at Qlik, enterprise agentic adoption is in its earliest stages. Customers are experimenting with low-risk use cases in controlled environments, but production-grade agents making real decisions in real business processes are rare. Data quality is the primary blocker. Gartner projects that by 2027, 70% of organizations will adopt modern data quality solutions to support AI initiatives.

What is Qlik’s Trust Score for AI?

Qlik’s Trust Score for AI is a quantifiable measure of whether data is valid, up to date, and representative before it reaches an AI agent or a large language model. It scores data across dimensions including diversity, timeliness, and accuracy, giving decision-makers visibility into data reliability rather than requiring them to take data quality on faith. Qlik has been named a Leader in the Gartner Magic Quadrant for Augmented Data Quality Solutions for seven consecutive years.

What does “dashboards are dead” mean?

Brendan Grady’s declaration that “dashboards are dead” refers to dashboards as a destination, not the data or insights within them. The traditional model of going to a dashboard to draw your own conclusions is being replaced by AI-powered interfaces that proactively recommend actions. Qlik calls this shift decision intelligence. Grady described his own workflow changing from “give me information about my business” to “tell me about my business and what you think I should do.”

What is the Qlik MCP server?

The Qlik MCP server implements the open Model Context Protocol, allowing third-party AI assistants such as Anthropic Claude, Google Gemini, and ChatGPT to access Qlik’s analytical capabilities, with built-in governance and audit trails. Launched in February 2026, it reflects Qlik’s bet on interoperability over platform lock-in. Grady reported that roughly 50% of users now access Qlik’s agentic capabilities through MCP rather than Qlik’s own interface.

What should data professionals do to prepare for the agentic AI era?

Brendan Grady advises data professionals to lean into what they already know best: the data itself. His recommendation is to become data product owners who serve as trusted guides as organizations adopt agentic AI. The people who understand data quality, business context, and organizational nuance will be indispensable because these capabilities are not ones AI agents can replicate on their own.

Podcast highlights

- [0:00] Introduction and welcome at Qlik Connect 2026

- [1:14] Brendan’s first job: Sound of Music tour guide in Salzburg

- [2:04] Lessons learned from the early analytics era

- [3:32] Why data quality has never been fixed

- [4:46] Consequence management in the agentic era

- [6:08] Where enterprise agentic adoption actually stands

- [7:46] Future-proofing against LLM shifts

- [8:24] The analytics engine and unknown unknowns

- [10:29] Structured vs. unstructured data convergence

- [12:04] Hallucinations and the trust problem

- [15:30] Decision intelligence and “dashboards are dead”

- [18:05] Brain outsourcing and the MIT cognitive debt study

- [21:57] MCP server and open standards

- [23:54] Key themes for Qlik in 2026: trust, context, flexibility

- [26:12] Advice for data professionals

- [28:15] Does AI expand the aperture for who can participate in analytics?

About David Sweenor

David Sweenor is the founder and host of the Data Faces Podcast, where he talks with the people who are making data, analytics, AI, and marketing work in the real world. He is also the founder of TinyTechGuides and a recognized top 25 analytics thought leader and international speaker who specializes in practical business applications of artificial intelligence and advanced analytics.

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence. David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Books

- Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

- Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

- The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

- The CIO’s Guide to Adopting Generative AI: Five Keys to Success

- Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

- The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn.

^[1]Sweenor, David. “Beyond the AI Hype: What 20% of Companies Get Right.” TinyTechGuides, February 11, 2025. https://tinytechguides.com/blog/beyond-the-ai-hype-what-20-of-companies-get-right/

^[2]Gartner. “Lack of AI-Ready Data Puts AI Projects at Risk.” Gartner Newsroom, February 26, 2025. https://www.gartner.com/en/newsroom/press-releases/2025-02-26-lack-of-ai-ready-data-puts-ai-projects-at-risk

^[3]Sweenor, David. “From ‘AI-Ready’ to AI Reality: Why Actionable Data Strategies Beat Endless Planning.” TinyTechGuides, June 3, 2025. https://tinytechguides.com/blog/from-ai-ready-to-ai-reality-shane-murray-on-data-trust-and-why-action-beats-planning/

^[4]Qlik. “Qlik Releases Trust Score for AI in Qlik Talend Cloud.” Qlik Press Release. https://www.qlik.com/us/news/company/press-room/press-releases/qlik-releases-trust-score-for-ai-in-qlik-talend-cloud

^[5]Qlik. “Qlik Named a Leader in the 2026 Gartner Magic Quadrant for Augmented Data Quality Solutions.” Qlik Press Release, 2026. https://www.qlik.com/us/news/company/press-room/press-releases/qlik-named-a-leader-in-the-2026-gartner-magic-quadrant-for-augmented-data-quality-solutions

^[6]Qlik. “Qlik Brings Agentic Analytics to General Availability and Launches MCP Server for Third-Party Assistants.” Qlik Press Release, February 10, 2026. https://www.qlik.com/us/news/company/press-room/press-releases/qlik-brings-agentic-analytics-to-general-availability-and-launches-mcp-server-for-third-party-assistants

^[7]MIT Media Lab. “Your Brain on ChatGPT: Accumulation of Cognitive Debt When Using an AI Assistant for Essay Writing Task.” MIT Media Lab, 2025. https://www.media.mit.edu/publications/your-brain-on-chatgpt/

On the Data Faces Podcast, I usually interview someone with a whole face. For this bonus segment, I made an exception.

Scott Taylor, the Data Whisperer, is known for his work in data management consulting and storytelling. He’s also the creator of Data Puppets, a satirical puppet series that uses humor to expose the enterprise data problems that executives resist hearing about directly. In Episode 35 of the Data Faces Podcast, Scott and I had a serious conversation about why data leaders keep losing the room and how storytelling wins it back. Then we let one of his puppet characters take the mic.

The character’s name is A-Eye. Not “an AI.” As the puppet put it, “I choose my own indefinite article. Personal branding is important.”

A-Eye had just returned from the Gartner Data & Analytics Summit. The report from the show floor was not subtle.

“AI was everywhere. Regular AI, gen AI, agentic AI, autonomous AI, in the loop AI, out of the loop AI, trapped in the loop AI. Some vendors were basing their whole future on it, and two years ago they couldn’t even spell AI.”

The agents impressed him most. “Agents writing code, agents reviewing code, deploying code, and then apologizing for the code. It’s a total system.”

I asked about data quality. A-Eye was unmoved. “They’ve been whining about data quality ever since there was data. If it was that important, would it have been solved by now?”

And governance? “AI is the Ozempic for data governance, baby. Your data never looked so good.”

The segment wrapped with A-Eye reworking Old MacDonald into a data anthem. I was asked to sing along. I did. I shouldn’t have.

Scott’s Data Puppets work because a puppet can say things that would come off as harsh from a human consultant. The CDO (Chief Dog Officer), IT Bee (who speaks only in buzzwords), and the Cat Sultant from Meow-kinsey have all become tools that data teams use in their own presentations to show leadership how the data team sounds to the business side. The satire lands because it’s uncomfortably accurate.

Watch the full conversation with Scott Taylor on the Data Faces Podcast.

Data leaders have been pitching “data quality” to executives for decades. For just as long, executives have nodded politely, approved a fraction of the requested budget, and moved on to whatever initiative sounds more exciting. Gartner estimates that poor data quality costs the average enterprise $12.9 to $15 million per year, yet data leaders still struggle to connect that cost to the language executives actually use.^[1]

“I can boil my entire data philosophy down to those three words: truth before meaning. You got to determine the truth in your data before you derive any kind of meaning out of it. It’s not chicken or egg here. This is egg and omelet.”

— Scott Taylor, Founder, MetaMeta Consulting

On Episode 34 of the Data Faces Podcast, I sat down with Scott Taylor to talk about why data management keeps getting sidelined and what data leaders can do about it. Scott has spent 30 years in the data space and now runs MetaMeta Consulting, where he helps organizations craft business-accessible narratives about data management. His central argument is that you have to establish truth in your data before you try to derive any meaning from it. Getting there requires storytelling, a skill that most data practitioners were never trained in.

About Scott Taylor

Scott Taylor is the founder of MetaMeta Consulting and is known across the data industry as “the Data Whisperer.” He has spent 30 years in the data space, including 25 years in corporate roles before becoming a full-time content creator, speaker, and consultant. Scott is also the creator of Data Puppets, a satirical puppet series that uses humor to expose common enterprise data problems, and the author of Telling Your Data Story. In our conversation on Episode 34 of the Data Faces Podcast, we discuss:

- Why “truth before meaning” is the foundational principle for every data initiative

- How data leaders can craft a one-sentence pitch that resonates with a skeptical CFO

- The 3V framework for data storytelling: Vocabulary, Voice, and Vision

- Why the vendor landscape at Gartner D&A looked “horrifyingly consistent”

- How Data Puppets uses satire to expose organizational dysfunction that executives resist hearing directly

Truth before meaning: egg and omelet, not chicken and egg

Every major trend in the data space follows the same cycle. A new technology arrives with enormous promise; organizations rush to adopt it; and at some point, someone in the room realizes that none of it works unless the underlying data is in good shape. Scott has watched this play out with big data, data mesh, data fabric, and now agentic AI over three decades. He frames the challenge with a phrase he has distilled into the fewest possible words: truth before meaning.

The “truth” part refers to the foundational work that most organizations know they need to do but struggle to prioritize: master data, reference data, metadata, MDM, data governance, and all the structural activities that ensure your data is curated, trusted, and fit for purpose. The “meaning” part is everything that gets the budget and the boardroom attention, from BI dashboards and AI models to analytics platforms and the latest agentic AI initiative.

Organizations consistently skip truth and jump straight to meaning. They invest in AI without verifying that their customer master is reliable. They build analytics on hierarchical structures that don’t hold up across departments.^[2] Research from IBM and MIT Sloan Management Review suggests that companies lose 15 to 25 percent of revenue because of poor data quality, and that cost only grows as organizations scale their AI investments without fixing the underlying data.^[3]

Scott illustrates the point with a supermarket example. Every time you take a product off a shelf and scan it at the register, that beep is a confirmation of truth. The system knows exactly what that product is, how it’s priced, and how it’s tracked. The challenge for enterprises is building that same level of confidence across hundreds of systems and millions of records.

“If the pitch that we need better data quality worked, then I wouldn’t be on your show, because people would be doing it. It would have been done. It wouldn’t be something that we’re still talking about.”

— Scott Taylor, Founder, MetaMeta Consulting

Why data leaders lose the room (and how storytelling wins it back)

I spent the first half of my career as a data practitioner and the second half in product marketing, so I’ve seen this from both sides. Data professionals are trained in hard skills, and nobody starts their data career with a course on how to pitch metadata management to a CFO. Marketing and sales teams learn to tell stories because their results depend on it, but data leaders tend to explain the mechanics first, walking into a meeting to describe the technical approach and losing their audience before they ever get to the business case.

“Data people love to get technical. They love to explain how it’s going to get done. And you just lose the business folks right away. A CEO, if you want money from them, they don’t care how it’s done until they understand why it’s important to the organization.”

— Scott Taylor, Founder, MetaMeta Consulting

Scott advises flipping the sequence. Every business leader has stated goals around growth, market expansion, customer experience, or operational efficiency. The data opportunities are already embedded in those objectives. If you listen to what your business leaders say they want to accomplish, you will find where the discipline fits. Their conversations focus on customers, brands, and markets, and all of those have a data element to them.

The 3V framework: Vocabulary, Voice, and Vision

Scott structures his advice around a framework he calls the 3V of data storytelling for data management. Most data practitioners already know the 3V of big data, and the parallel is deliberate.

Vocabulary. Get the words right. The language you use to describe data management to a CFO should be different from the language you use with your data engineering team. Terms like “master data management” and “reference data governance” mean nothing to someone whose primary concern is revenue growth or margin improvement. Scott recommends using words like strength, structure, and foundation rather than quality, because quality can feel subjective and emotional, almost like a complaint when what you need is a business case.

Voice. Everyone involved in making the case for data management needs to tell a consistent story. If the data team, the IT team, and the business analysts are all framing the problem differently, the message gets diluted before it reaches the people who control the budget. Scott calls this harmonizing to a common voice across the organization.

Vision. Connect every data activity to the strategic intentions of the enterprise. The pitch becomes impossible to ignore when you frame data investment as the enabler of business outcomes the leadership team has already committed to achieving.

“You’ve got to connect those dots between why we need metadata management in the context layer to the CEO’s initiative of expanding to new markets and becoming better partners with our customers.”

— Scott Taylor, Founder, MetaMeta Consulting

AI is not the Ozempic for data governance

At the Gartner D&A Summit in Orlando, Scott and I both noticed the same thing on the show floor. The vendor messaging was, in Scott’s words, “horrifyingly consistent.” Nearly every booth was leading with agentic AI, AI-native architecture, and context layers. Scott’s tongue-in-cheek response was classic Scott. As people posted about vendor after vendor emphasizing “context,” he started commenting on their posts with the same line: “context is the new oil.”

This hype cycle mania repeats every few years, from data mesh and data fabric three years ago to big data before that, and the cycle always ends with organizations realizing the new thing doesn’t work without solid data management underneath it.^[4] Scott’s colleague Malcolm Hawker coined a name for it: the “semantic pedantic cycle.”

The way Scott sees it, the belief that AI will solve the data management problem on its own is the latest version of this thinking. He calls this “AI is the Ozempic for data governance,” a line that got plenty of laughs at Gartner and in our bonus Data Puppets segment. AI can assist with certain data management tasks, and the organizational discipline of establishing truth in your data before deriving meaning from it still requires human leadership and commitment.

Data Puppets: using satire to say what executives need to hear

Think of it as Dilbert for the data world, except with puppets. The cast includes the CDO (Chief Dog Officer), his sidekick ITB who speaks exclusively in buzzwords, and a “Cat Sultant” from Meow-kinsey whose primary initiative is to generate more billing. Just as Scott Adams captured the absurdity of corporate life in ways that employees pinned to their cubicle walls, Scott Taylor captures the absurdity of enterprise data management in ways that data teams share in their Slack channels.

What started as a collection of data jokes has turned into a communication tool that Scott didn’t anticipate. People use the episodes in internal presentations to illustrate how the data team sounds to the business side. A Chief Dog Officer can say things that would come off as harsh from a human consultant, and people laugh first and then recognize the pattern in their own organization.

In the bonus Data Puppets segment at the end of our recording, Scott introduced A-Eye, a puppet character who attended the Gartner D&A Summit and had opinions about everything. When asked about data quality, A-Eye’s response captured an attitude that data leaders encounter constantly: “They’ve been whining about data quality ever since there was data. If it was that important, it would have been solved by now.”

“The number one reaction I got was, ‘this is just like my organization.’ People were really taking it seriously. They were like, ‘I showed this to my team to show this is how we sound to the business side.’”

— Scott Taylor, Founder, MetaMeta Consulting

Next steps

Scott’s approach offers a practical starting point for data leaders who are struggling to get executive support for foundational data work. The investment is in changing the conversation, which costs nothing beyond the willingness to rethink how you communicate.

- Craft your one-sentence pitch. Distill why data management matters to your organization into the fewest possible words. Frame it as a business statement that connects to what leadership has already said they want to accomplish, because a technical explanation won’t land. If you can’t say it in one sentence, you haven’t refined it enough.

- Audit your storytelling sequence. Are you leading with how (the technical approach) or why (the business impact)? If your presentations start with architecture diagrams and technology stacks, consider flipping the order. Open with the business objective, show how data enables it, and save the technical details for the appendix.

- Apply the 3V framework. Review the vocabulary you’re using with executive stakeholders. Swap subjective terms like “data quality” for structural language like “data foundation” and “data trust.” Align your team to a common voice so the message doesn’t fragment across departments. Make sure every data initiative you propose connects to the organization’s stated strategic vision.

Listen to the full conversation with Scott Taylor on the Data Faces Podcast.

Based on insights from Scott Taylor, Founder at MetaMeta Consulting, featured on the Data Faces Podcast.

Share TinyTechGuides

Leave a comment

Frequently asked questions

What does “truth before meaning” mean in data management?

Truth before meaning is the principle that organizations must establish trustworthy, well-governed foundational data before attempting to derive business insights from it. The “truth” layer includes master data, reference data, metadata, and data governance. The “meaning” layer includes BI, analytics, AI, and everything that interprets data for business decisions. Scott Taylor argues that skipping the truth layer is the primary reason AI and analytics initiatives underperform. Gartner estimates that poor data quality costs the average enterprise $12.9 to $15 million per year.

What is the 3V framework for data storytelling?

The 3V framework is Scott Taylor’s approach to helping data leaders communicate with executive stakeholders. It stands for Vocabulary (choosing business language over technical jargon), Voice (aligning the entire data organization around a consistent narrative), and Vision (connecting every data initiative to the company’s strategic objectives). The framework is designed to shift data management conversations from technical how-to discussions to business-impact narratives that resonate with CFOs and CEOs.

Why do data leaders struggle to get executive buy-in for data management?

Data leaders are trained in hard skills and tend to lead presentations with technical approaches rather than business outcomes. Executives in marketing, sales, and the C-suite are better storytellers by training and practice. When data leaders explain master data management architecture before explaining how it connects to revenue growth or market expansion, they lose their audience. Scott Taylor recommends flipping the sequence and leading with why the work matters to the business before explaining how it gets done.

Can AI fix data quality problems on its own?

No. Scott Taylor calls the belief that AI can solve data management problems without organizational discipline “AI is the Ozempic for data governance.” While AI can assist with specific data management tasks, it cannot replace the foundational work of establishing trusted master data, standard hierarchies, and consistent taxonomies. Research from IBM and MIT Sloan Management Review suggests companies lose 15 to 25 percent of revenue from poor data quality, and that cost scales as AI investments grow without addressing the underlying data.

What are Data Puppets, and why do they matter for data management?

Data Puppets is a satirical puppet series created by Scott Taylor that uses humor to expose common enterprise data dysfunction. Characters include the CDO (Chief Dog Officer), a buzzword-fluent sidekick named ITB, and a consultant from “Meow-kinsey.” The series works as a communication tool because satire creates a layer of separation that lets audiences absorb uncomfortable truths about their own organizations. People use the episodes in internal presentations to illustrate how data teams sound to business stakeholders.

Podcast highlights

[0:06] Scott’s background as the Data Whisperer and 30 years in the data space

[3:59] Truth before meaning: Scott’s entire data philosophy in three words

[6:04] Why data truth isn’t philosophical, and the supermarket scanner example

[7:56] The importance of storytelling and why data practitioners aren’t trained in it

[10:27] Has AI changed the conversation about data management, or is it the same cycle?

[13:08] How vendors performed at the Gartner D&A Summit in Orlando

[16:27] “Context is the new oil” and the semantic pedantic cycle

[19:54] Crafting a one-sentence data management story for a skeptical CFO

[22:59] The 3V framework: Vocabulary, Voice, and Vision

[25:37] Data Puppets: how satire reveals organizational dysfunction

[31:48] Why humor helps executives hear truths they’d otherwise dismiss

[34:24] Where to find Scott Taylor and the Data Puppets

Bonus: Data Puppets segment — A-Eye attends the Gartner D&A Summit

About David Sweenor

David Sweenor is the founder and host of the Data Faces podcast, where he talks with the people who are making data, analytics, AI, and marketing work in the real world. He is also the founder of TinyTechGuides and a recognized top 25 analytics thought leader and international speaker who specializes in practical business applications of artificial intelligence and advanced analytics.

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence. David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Books

- Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

- Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

- The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

- The CIO’s Guide to Adopting Generative AI: Five Keys to Success

- Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

- The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn.

^[1]Gartner. “How to Improve Your Data Quality.” Gartner, 2021. https://www.gartner.com/smarterwithgartner/how-to-improve-your-data-quality.

^[2]David Sweenor. “AI in 2025: Why 90% of Gen AI Projects Will Fail.” TinyTechGuides, March 22, 2025.

https://insights.tinytechguides.com/p/ai-in-2025-why-90-of-gen-ai-projects

^[3]IBM. “The True Cost of Poor Data Quality.” IBM Think, 2024. https://www.ibm.com/think/insights/cost-of-poor-data-quality.

^[4]David Sweenor. “AI in 2025: Why 90% of Gen AI Projects Will Fail.” TinyTechGuides, March 22, 2025.

https://insights.tinytechguides.com/p/ai-in-2025-why-90-of-gen-ai-projects

.

I spend most of my time consulting with organizations that are trying to figure out what to do with AI, and teaching is a big part of that work. I’ve also conducted more than 35 interviews on the Data Faces Podcast with data leaders, practitioners, and technology executives. The question I hear in every engagement and nearly every episode is the same: how do I know the output is right? When a chatbot gives a questionable answer, someone catches it and moves on. An autonomous agent, on the other hand, might already be three decisions downstream before anyone notices the answer was wrong.

At the Gartner Data & Analytics Summit in Orlando, I sat down with Conor Jensen for an on-location episode of the Data Faces Podcast. Conor is the Global Field CDO at Dataiku, a data science and machine learning platform used by enterprise organizations to build, deploy, and manage AI projects. It’s a role shaped by an unusual path. He purchased Dataiku as a customer about ten years ago, spent seven years on the other side of the table, and now helps organizations avoid the mistakes he already made. He’d just come off Dataiku’s biggest product launch in the company’s 13-year history, and one observation from our conversation captured exactly what I’ve been hearing from clients.

“A far more dangerous thing than an agent that breaks is an agent that’s still functioning and giving the wrong answers.” — Conor Jensen, Global Field CDO, Dataiku

According to Gartner, only 6% of organizations have AI agents in production today, while 53% are still in exploration mode.^[1] The organizations racing to build agents have largely skipped the question of whether the ones they already have are performing.

About Conor Jensen

- Conor Jensen is the Global Field CDO at Dataiku. He purchased Dataiku as a customer about ten years ago, joined the company seven years later, and now helps organizations develop AI strategy and operational plans to get the most out of the platform. Before Dataiku, he worked as a data scientist and analytics leader.

- Key topics discussed: Dataiku CoBuild and vibe coding for data pipelines, Reasoning Systems for multi-step autonomous decisions, the Agent Management Platform for cross-platform observability, getting AI from pilot to production, and why perfect data is never coming

Everyone’s building agents, nobody’s solved production

I asked Conor why so many AI projects stall between prototype and production. He didn’t point to a single bottleneck. He described a pile of them that keeps growing.

“We haven’t solved any of that yet as an industry. We just keep putting more in the backpack.” — Conor Jensen, Global Field CDO, Dataiku

MLOps was supposed to get machine learning models into production. Then came LLMOps for large language models. Now the industry is talking about AgentOps. Each layer adds new complexity without resolving the one that came before it. Conor sees three barriers that keep organizations stuck. Deployment architecture is one, where something that works on a laptop or in a dev environment falls apart on the way to production. Organizational dynamics are another, including governance, trust, and change management, which he considers harder than any technical challenge. And then there’s data readiness, where teams wait for perfect data that will never arrive.

Gartner’s research reinforces how high the stakes are. Rita Sallam estimates that 70% of agentic AI use cases will fail to deliver expected value due to underinvestment in necessary foundations.^[2] Data availability and quality remain the number one barrier to AI implementation, cited by 30% of data management leaders.^[3] Gartner analyst Sarah Turkaly reinforced the point at the summit: “Data governance will be the single point of failure for organizations’ AI ambitions.”^[4]

Dataiku’s biggest launch targets every layer of the problem

The opening keynote from Adam Ronthal and Georgia O’Callaghan set the tone for the summit by framing AI value around three returns: return on intelligence, return on integration, and return on individuals.^[5] Dataiku positioned its announcement around that same framework, branding the launch as “The Platform for AI Success.” Conor walked me through three new products that Dataiku announced at the summit, each one targeting a different layer of the production problem.

Dataiku CoBuild brings vibe coding into the data platform, but the comparison to building a web app breaks down quickly. With a web app, you click a button and the page loads or it doesn’t. With a data pipeline, you get summary statistics and a model, but verifying the answer requires a level of inspection that 2,000 lines of generated Python won’t give you. CoBuild takes that generated code and renders it as visual workflows you can step through, edit, and validate. Conor, a data scientist himself, was candid about why this matters.

“Out of 2,000 lines of Python and a machine learning project, there’s probably like 40 that are what’s really, really important. The rest of it is, yeah, okay, did you pull the right data?” — Conor Jensen, Global Field CDO, Dataiku

CoBuild abstracts the boilerplate so you can focus on the 40 lines that determine whether the output is trustworthy. It launches in June 2026.

Reasoning Systems tackle a different gap. Conor used the example of a supply chain analyst who today pulls data from five different systems, consults with other teams, and makes a judgment call. Reasoning Systems layer process flows and context on top of data sources, then give an agent the ability to walk through the entire sequence. The key difference from RPA is that not every step is deterministic. Some require the agent to self-correct or stop entirely. Dataiku is building these for targeted use cases in specific industries rather than trying to solve everything at once.

The product Conor said he’s personally most excited about is the Agent Management Platform. Fifty-four percent of organizations are exploring or deploying goal-driven AI agents, according to Gartner.^[6] The question most CIOs should be asking is straightforward: how many agents do I have in production across all of my systems? With agents being built and deployed on Databricks, Salesforce, and dozens of other platforms alongside Dataiku, that question is hard to answer today.

“How do I manage all of my agents across my infrastructure, wherever they’ve been deployed? How do I make sure I know that they’re performing, not just functioning, but performing?” — Conor Jensen, Global Field CDO, Dataiku

Monitoring whether an agent is running is table stakes. You can do that with an API bus. The Agent Management Platform goes further by adding performance management, a semantic layer, and contextual understanding across every environment where agents are deployed. It evaluates whether agents are delivering the right business results across eight, ten, or twenty different systems. It goes GA in September 2026 and does not require being a Dataiku customer.

Conor had practical advice for organizations that feel stuck waiting for perfect data or an industry standard to emerge.

“News flash. There’s no such thing as perfect data, never will be. You have to just get moving.” — Conor Jensen, Global Field CDO, Dataiku

Only 12% of D&A leaders say they are fully prepared to carry out their mandate, according to Gartner’s 2026 CDAO survey.^[7] Conor’s point is that treating full readiness as a prerequisite for action is its own form of failure.

Production is the starting line

Conor Jensen has seen the Dataiku platform from both sides over the past decade, and that practitioner-turned-vendor perspective came through in every answer he gave. The industry has spent years talking about getting AI to production. The conversation at Gartner this year made clear that production is only the starting line. The harder work is knowing what happens after you deploy, and most organizations have no way to answer that question across their agent portfolio today.

The next time someone on your team proposes building a new agent, ask a different question first. Do you know how the ones you already have are performing?

Listen to the full conversation with Conor Jensen on the Data Faces Podcast.

Based on insights from Conor Jensen, Global Field CDO at Dataiku, featured on the Data Faces Podcast.

Share

Share TinyTechGuides

Leave a comment

Frequently asked questions

What is Dataiku’s Agent Management Platform? Dataiku’s Agent Management Platform provides cross-platform observability and performance management for AI agents deployed across any system, including those built outside of Dataiku. It goes beyond uptime monitoring by evaluating whether agents are delivering correct business results. The platform adds a semantic layer and contextual understanding so organizations can assess agent performance across eight, ten, or twenty different environments from a single view. It is scheduled for general availability in September 2026.

How is vibe coding a data pipeline different from vibe coding a web app? With a web app, you can visually confirm whether it works by clicking a button and seeing the result. With a data pipeline, AI generates thousands of lines of code that produce summary statistics and a model, but there is no simple way to verify the answer is correct. Dataiku CoBuild addresses this by rendering generated code as visual workflows that users can step through, edit, and validate rather than reading through 2,000 lines of Python.

What are Dataiku Reasoning Systems? Reasoning Systems layer process flows and business context on top of data sources to enable multi-step autonomous decisions. Unlike RPA, where every step is deterministic, Reasoning Systems allow agents to self-correct or stop when results fall outside expected parameters. Dataiku is building these for targeted use cases in specific industries, starting with manufacturing operations, with supply chain and financial risk scheduled for later in 2026.

Why do most agentic AI use cases fail? Gartner estimates that 70% of agentic AI use cases will fail to deliver expected value due to underinvestment in necessary foundations. The top barrier to AI implementation is data availability and quality, cited by 30% of data management leaders. Organizations also struggle with deployment architecture, governance, and change management. Gartner analyst Sarah Turkaly warned that data governance will be the single point of failure for organizations’ AI ambitions.

How many organizations have AI agents in production? According to Gartner research from January 2025 surveying 3,412 respondents, only 6% of organizations have AI agents in production. Fifty-three percent are still in exploration mode, and 25% are piloting. Fifty-four percent of organizations are exploring or deploying goal-driven AI agents, but most cannot answer how many agents they have running across their infrastructure or whether those agents are delivering correct results.

Podcast highlights

[0:00] Introduction at the Gartner D&A Summit and Dataiku overview

[1:27] Three new product announcements: CoBuild, Reasoning Systems, Agent Management Platform

[2:49] Dataiku’s evolution in the age of Gen AI

[3:30] Why AI projects stay stuck in pilot purgatory

[5:30] Deployment architecture that works from dev to production

[7:00] CoBuild, vibe coding, and why data pipelines are different from web apps

[8:26] Why even data scientists need better coding practices

[9:32] Reasoning Systems and autonomous multi-step decisions

[11:01] Agent Management Platform and cross-platform observability

[13:00] Monitoring vs. performance management for agents

[15:00] Opening the gates with governance and guardrails

[17:00] GA timeline, availability, and closing

About David Sweenor

David Sweenor is an AI advisor, author, and the founder of TinyTechGuides. He spent the first half of his career as a practitioner at IBM, building data warehouses and running predictive models, and the second half in product marketing leadership at SAS, Dell, TIBCO, Alteryx, and Alation. He advises Fortune 500 companies on AI strategy, data governance, and go-to-market planning, and hosts the Data Faces Podcast, where he interviews the leaders, practitioners, and technologists shaping the future of data and AI.

Books

- Artificial Intelligence

- Generative AI Business Applications

- The Generative AI Practitioner’s Guide

- The CIO’s Guide to Adopting Generative AI

- Modern B2B Marketing

- The PMM’s Prompt Playbook

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn.

^[1]Chandrasekaran, Arun. “Navigating the AI Agent Landscape: A Strategic Guide for IT Leaders.” Gartner D&A Summit 2026, March 2026.

^[2]Sallam, Rita. “How to Calculate the Value and Cost of AI Agents.” Gartner D&A Summit 2026, March 2026.

^[3]Ramakrishnan, Ramke. “How Is Agentic AI Impacting and Disrupting Your Data Management Discipline?” Gartner D&A Summit 2026, March 2026.

^[4]Turkaly, Sarah. “The Future of D&A Governance.” Gartner D&A Summit 2026, March 2026.

^[5]Ronthal, Adam and Georgia O’Callaghan. “Navigate AI on Your Data & Analytics Journey to Value.” Gartner D&A Summit 2026 Opening Keynote, March 9, 2026.

^[6]Ramakrishnan, Ramke. “How Is Agentic AI Impacting and Disrupting Your Data Management Discipline?” Gartner D&A Summit 2026, March 2026.

^[7]Gabbard, Michael. “Signature Series: State of D&A 2026.” Gartner D&A Summit 2026, March 2026.

The spreadsheet might still be the most widely used data catalog on the market. That’s not a joke. It’s a finding from Stewart Bond, Research VP at IDC, who has spent the past decade studying how companies manage intelligence about their data. When I sat down with Stewart on the Data Faces podcast, he pointed out that every survey he runs surfaces the same contradiction. Organizations rank data quality as their top AI concern, yet they fail to invest in the one technology category designed to address it.

The frustrating part is that the data teams usually know exactly what the problem is. They flag quality and governance issues, but the budget continues to flow towards AI model development and agents instead. Stewart has been tracking this gap longer than most, and his perspective on how data intelligence evolved from an analyst’s shorthand into a global market category offers a useful lens for understanding why the gap persists.

“One of the biggest challenges organizations have is managing the intelligence about their data. Data catalogs, business glossaries, data lineage, all that stuff is so important now as we get into AI. And yet, their top investment categories are not on data catalogs.” — Stewart Bond, Research VP, IDC

About Stewart Bond

Stewart Bond is a Research VP at IDC, where he leads the data intelligence and data integration software research practice. His career spans over 30 years in IT, including a decade as a certified IT architect at IBM before moving into industry analysis in 2011. Outside of work, Stewart is a competitive curler who came within one match of representing Ontario at a Canadian national championship. In our conversation on the Data Faces Podcast, we discuss:

• How Stewart coined the term “data intelligence” and watched it become a global market category

• The difference between intelligence about data and intelligence from data

• Why agentic AI demands a shift-left approach to data quality

• What CDOs are most concerned about and where they’re under-investing

How one research note became a market category

Stewart joined IDC in 2015 and inherited a research area covering data integration and data access, which at the time included eight sub-markets like metadata management, data quality, and master data. A conversation with ASG Technologies introduced him to their term “enterprise data intelligence.” Stewart saw something useful in the phrase but dropped the “enterprise” qualifier. Data intelligence, as a simpler label, stuck.

The real momentum came in 2018, when GDPR was about to take effect. Enterprise data leaders started calling Stewart with the same question. “Where can I buy a data governance solution?” His response surprised them. You can’t buy governance. Data governance is an organizational discipline that requires people, processes, and accountability. What you can buy is data intelligence technology, the tools that tell you everything you need to know about your data so you can govern it.

“I had a lot of end-user clients calling me and saying, ‘Where can I buy a data governance solution?’ And I just kind of laughed, because data governance isn’t a technology solution.” — Stewart Bond, Research VP, IDC

Stewart framed this through his 5 W’s of data. Who is using it? How is it being used? Where does it live? What does it mean? Why do you even have it? How long do you have to keep it? These questions form the foundation of effective data governance, and answering them requires technology that most organizations still haven’t fully invested in.^[1]

Intelligence about data vs. intelligence from data

The term spread faster than Stewart expected. Collibra became “the data intelligence company.” Erwin (now Quest) adopted it for their data catalog. Alation started using it in 2020, after learning the phrase wasn’t a Collibra trademark but an industry-level concept. Informatica wove it into their intelligent data platform messaging. Then, in late 2023, Databricks made a major push with its own version of data intelligence.

The Databricks definition, however, expanded the original meaning. Stewart had always treated data intelligence as intelligence about data. What is this data, where did it come from, who uses it, and how good is it? Databricks extended the concept to include intelligence from data, using the metadata and context layer to generate smarter analytics and AI outcomes from the data itself. The distinction matters because it changes what organizations expect from the category and how they evaluate platforms.

Dave Kellogg was serving as acting CMO at Alation when he first explored the term’s origins with Stewart. After the Databricks announcement, Kellogg reached out with a direct assessment. “I think you did it. I think you created a new market category.” Last year, IBM confirmed the trend by rolling its entire portfolio of data cataloging, quality, lineage, and observability products into IBM watsonx Data Intelligence. IBM’s product leadership told Stewart the renaming was a direct result of his work and the broader market momentum he helped create.

“I’d always treated data intelligence as intelligence about the data. I’d say Databricks has extended it to intelligence from the data, getting more into the case of leveraging that intelligence about the data to make sure you’re using the data intelligently.” — Stewart Bond, Research VP, IDC

Agents can’t wait for clean data

The shift to agentic AI fundamentally changes how organizations need to approach data quality. Traditional analytics workflows gave organizations a buffer. Data moved through batch processes, giving teams time to spot anomalies and intervene before a bad number reached a dashboard. Autonomous agents don’t offer that luxury. An agent monitoring a change data capture stream sees a new order event and starts fulfilling it on the spot. If the data in that event is wrong, the agent acts on it before anyone has a chance to review it.

Stewart describes this as the “shift left” imperative. Data quality, privacy, and integrity all need to move as close to the source as possible, because once data enters the agentic pipeline, there is no batch window to clean it up. Deloitte flagged this as one of four critical data quality challenges for AI, finding that companies building agentic systems need quality controls embedded at the point of data creation, not applied after the fact.^[2]

“You’d better make sure the data in that order event is good and that it’s a real and reliable order event. You may have heard the term shift left. Your data quality, your data privacy, your data integrity all need to be as close to the source as possible.” — Stewart Bond, Research VP, IDC

The challenge extends beyond structured data. Stewart raised a question that most organizations still haven’t answered well. What do you do about the unstructured data that makes up the bulk of enterprise information? Every organization has countless versions of the same PowerPoint file, thousands of PDFs, and documents that LLMs are eager to ingest. Some vendors are starting to crack this problem. Shelf.io, for example, has developed methods to assess the quality of unstructured documents, a capability that seemed impossible just a few years ago.

The broader issue remains, though. Most organizations lack the data context needed to determine whether their unstructured data is safe to use, let alone high-quality.^[3] Stewart sees agentic AI as part of the eventual solution. Agents that pre-populate data catalogs and reduce the manual burden on data stewards could finally solve the adoption problem that has held these tools back for years. But that future depends on investing in the foundation today.

The investment gap CDOs can’t ignore

Stewart runs an annual survey of the Office of the Chief Data Officer, and the results tell a consistent story. When you ask CDOs what their biggest organizational concern is, skills top the list. They struggle to find people who can do the work. The second concern is managing expectations around what AI can deliver, not just within their own teams, but across the C-suite, where leadership is under pressure to show results quickly and often treats AI as a magic bullet.

“Their top investment categories are not on data catalogs. Back to the spreadsheet might still be the most widely used data catalog on the market. I don’t have data to prove that, but anecdotally, that could be the case.” — Stewart Bond, Research VP, IDC

What makes this frustrating is that CDOs now have more influence over IT spending than ever before. IDC predicted in 2024 that chief data officers would gain significantly more budget authority by 2025, driven by the fact that every major AI concern in enterprise surveys points to data: quality, correctness, privacy, and security. CDOs are accountable for all of it. Deloitte’s 2025 CDO Survey tells a similar story. These leaders are increasingly expected to demonstrate direct business impact from their data programs, even as their organizations resist the investments required to achieve it.^[4]

And yet, when Stewart looks at where enterprises are actually putting their money, the top investment categories are not data catalogs or data quality tools. The data lineage, metadata management, and business glossary capabilities that form the backbone of data intelligence remain underfunded, even as AI programs depend on them.^[5] That spreadsheet Stewart mentioned at the top of our conversation? For many organizations, it is still doing the job that a proper data catalog should be doing.

You’ll never score 100%

Stewart closed our conversation with an insight he picked up years before he even joined IDC. A life insurance company told him they had finally accepted that their data would never be 100% clean. Instead of chasing perfection, they started measuring how clean or dirty their data was and feeding that score into their calculations. Their actuaries knew how to work with uncertainty. They just needed the number.

Data intelligence doesn’t promise perfect data. It gives you a clear picture of how much you can trust what you have. Organizations that know the quality of their data before it enters an AI pipeline avoid the costly cycle of debugging outputs that were doomed from the start. A data quality score of 75 means something different from a score of 95, and both are more useful than no score at all. When that score travels alongside the data into an AI model or an autonomous agent, the organization can make informed decisions about how much confidence to place in the output.

Start with Stewart’s 5 W’s. Audit how your organization currently tracks who uses its data, where it lives, and how trustworthy it is. If the answer is a spreadsheet, you have your business case.

The spreadsheet is still winning. It doesn’t have to be.

Listen to the full conversation with Stewart Bond on the Data Faces Podcast.

Based on insights from Stewart Bond, Research VP at IDC, featured on the Data Faces Podcast.

Share

Leave a comment

Frequently asked questions

What is data intelligence?

Data intelligence is the category of technology that provides intelligence about your data. It encompasses data catalogs, business glossaries, data lineage, data quality, and metadata management. Stewart Bond, Research VP at IDC, coined the term to describe the tools that answer foundational questions about data, including who uses it, where it lives, what it means, and how trustworthy it is. More recently, vendors like Databricks have expanded the definition to also include intelligence from data, using that context layer to improve analytics and AI outcomes.

How does data intelligence differ from data governance?

Data governance is an organizational discipline that requires people, processes, and accountability. Data intelligence is the technology that supports it. You cannot buy a data governance solution, but you can invest in data intelligence tools that tell you everything you need to know about your data so you can govern it. Organizations that try to solve governance with technology alone tend to fail, according to IDC’s Stewart Bond.

Why does agentic AI require a shift-left approach to data quality?

Traditional analytics workflows gave teams time to spot and fix data issues in batch processes before results appeared on the dashboard. Autonomous AI agents operate in real time and act on data the moment they receive it, with no batch window to clean things up. This means data quality, privacy, and integrity controls need to move as close to the data source as possible. Deloitte identified this as one of four critical data quality challenges for organizations building agentic AI systems.

What are CDOs most concerned about in 2025?

According to IDC’s annual survey of the Office of the Chief Data Officer, skills gaps rank as the top concern. CDOs struggle to find qualified people to do the work. The second biggest concern is managing leadership expectations around what AI can realistically deliver. Despite growing influence over IT budgets, CDOs face a persistent disconnect between the data foundation AI requires and where their organizations actually invest.

Where are organizations under-investing in data intelligence?

IDC survey data show that the top enterprise investment categories are not data catalogs, data quality tools, or data lineage capabilities, even though managing data intelligence is one of the biggest challenges organizations report. Stewart Bond notes that the spreadsheet may still be the most widely used data catalog on the market, a sign that foundational data intelligence technology remains significantly under-funded relative to AI program spending.

Podcast highlights

[0:05] Introduction and Stewart’s background at IDC

[2:31] Stewart’s life outside work, competitive curling, and fishing

[5:00] The origin of the term “data intelligence” and the ASG Technologies connection

[6:44] GDPR drives demand for governance solutions, the 5 W’s of data

[8:15] Collibra, Erwin, Alation, and Informatica adopt the term

[10:00] Databricks expands the definition, Dave Kellogg’s “you created a category” moment

[14:00] IBM rebrands to watsonx Data Intelligence

[18:00] Intelligence about data vs. intelligence from data

[26:00] Agentic AI and the shift-left imperative for data quality

[29:00] Unstructured data quality and Shelf.io

[31:00] What CDOs are most concerned about in 2025

[35:00] Where organizations are under-investing in data intelligence

[36:40] Data quality will never be 100%, the life insurance anecdote

[38:00] Agentic AI and the future of data catalog adoption

About David Sweenor

David Sweenor is a Top 25 AI thought leader, six-time author, and founder of TinyTechGuides. He spent the first half of his career as a practitioner at IBM, building data warehouses and running predictive models, and the second half in product marketing leadership at SAS, Dell, Quest, TIBCO, Alteryx, and Alation. He hosts the Data Faces Podcast, where he talks with the people who are making data, analytics, and AI work in the real world.

Books

- Artificial Intelligence

- Generative AI Business Applications

- The Generative AI Practitioner’s Guide

- The CIO’s Guide to Adopting Generative AI

- Modern B2B Marketing

- The PMM’s Prompt Playbook

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn.

^[1]Sweenor, David. “Why the Biggest AI Enthusiasts Care Most About Governance.” TinyTechGuides, January 27, 2026. https://tinytechguides.com/blog/why-the-biggest-ai-enthusiasts-care-most-about-governance/

^[2]Deloitte. “Four Data and Model Quality Challenges for AI.” Deloitte AI Institute, 2025. https://www.deloitte.com/global/en/our-thinking/insights/topics/artificial-intelligence/ai-data-quality-challenges.html

^[3]Sweenor, David. “Your AI Doesn’t Have a Model Problem. It Has a Data Context Problem.” TinyTechGuides, February 24, 2026. https://tinytechguides.com/blog/your-ai-doesnt-have-a-model-problem-it-has-a-data-context-problem/

^[4]Deloitte UK. “CDO Survey 2025.” Deloitte United Kingdom, 2025. https://www.deloitte.com/uk/en/services/consulting/analysis/chief-data-officer-survey.html

^[5]Sweenor, David. “Data Lineage for AI: Why Truth Beats Hope in Banking.” TinyTechGuides, December 2, 2025. https://tinytechguides.com/blog/data-lineage-for-ai-why-dotrth-beats-hope-in-banking/

I walked the expo floor at the Gartner Data & Analytics Summit in Orlando expecting every conversation to be about AI, agents, and context layers. They were, and plenty of vendors had agent-washed their messaging overnight. The vendors were talking about the future, but the practitioners were pointing to something they already had.

The opening keynote set the tone. Adam Ronthal and Georgia O’Callaghan reported that four out of five organizations are now deploying AI, but only one in five will achieve their stated ROI.^[1] Governance, they argued, is a value accelerator and should be treated as one. With AI agents on every vendor’s booth and in nearly every session title, the question of how to govern autonomous systems had real urgency behind it.

I carried that framing into on-location interviews for the Data Faces Podcast with Philip Dutton, CEO and founder of Solidatus, Terrence Hedin, Data and Metadata Platform Director at the London Stock Exchange Group, and Caleb Watkins, Solutions Engineer at Solidatus. Three different roles, three different vantage points, and they all pointed to the same thing. The most valuable AI governance asset many organizations have is the data lineage and metadata infrastructure that their compliance teams built years ago. Solidatus, a data lineage and metadata management platform used by financial services and other regulated industries, served as the common thread across all three conversations.

From second-class citizen to strategic asset

I suggested to Terrence Hedin that before AI changed the conversation, lineage and metadata were treated as second-class citizens. He expanded on that.

“It has evolved. It is a first-class citizen,” Terrence said. “Every business requirement spec includes lineage at an element level. Every tech spec includes how you produce that lineage.”

At LSEG, lineage used to answer a narrow set of questions. Where did this data come from? Can we prove it to regulators? Those questions still matter, but Terrence described how LSEG now brings business metadata, technical metadata, and semantic layers together into a knowledge graph that serves the entire organization.

“We bring our business metadata, our technical metadata, our semantic layers, into a knowledge graph so we can build that true business context. That provides not only human benefit, but machine benefit as well.” — Terrence Hedin, Data and Metadata Platform Director, LSEG

LSEG now treats metadata as a data product, published to both internal teams and external customers. Not a theoretical data mesh exercise, but a commercial product. The governance infrastructure they built for compliance became the foundation for a revenue-generating line of business.

Gartner research supports this trajectory. In the session “Trust as the New Currency,” Guido De Simoni presented data showing that organizations with graduated trust models achieve 64% compliance success compared to 23% without them.^[2] The trust frameworks that organizations like LSEG built for regulators directly support AI readiness.

Caleb Watkins, a Solutions Engineer at Solidatus, showed me a related capability. Because Solidatus centralizes all data and metadata in one place, organizations can load their regulations as reference models and let the AI assistant evaluate compliance across their data landscape.

“We can train Solidatus up on those regulations, and then we can ask the assistant to assess your models for compliance with these different regulations to make sure that you’re meeting all of your objectives.” — Caleb Watkins, Solutions Engineer, Solidatus

Lineage is no longer just a record of where data came from. It’s becoming the system that evaluates whether your data meets the obligations attached to it.

Your compliance operating model already works for AI

I said something to Philip Dutton that surprised us both. “Something that was built for compliance is now incredibly useful for AI.”

Philip didn’t hesitate. Whether it’s an AI consuming data, a BI dashboard pulling reports, or another system sharing information across business lines, the obligations are the same. Purpose limitations, storage rules, and sharing boundaries all travel with the data.

“You don’t have to change your operating model for AI governance. You can use the same operating model that you’ve been using, which the organization knows, and it takes them a long time to get to know it and to feel comfortable with it. So this really gives you a nice accelerator.” — Philip Dutton, CEO and Founder, Solidatus

The program already exists, and your teams know how to run it. The organizational trust has been earned over years of practice. Rather than standing up a parallel AI governance function, extend the operating model you already have.

Gartner analyst Andrés García-Rodeja reinforced this point in the session “How to Build the Context Layer for Reliable AI Agents.” By 2028, he estimates, 60% of agentic analytics projects relying solely on the Model Context Protocol will fail due to the lack of a consistent semantic layer.^[3] The metadata and lineage infrastructure that compliance teams maintain is exactly the kind of semantic foundation AI agents need to operate reliably. For AI teams building production pipelines and deploying agents, the implication is direct: the semantic layer your models need may already exist in your governance program.

And that operating model is getting faster. Caleb walked me through code scanning with the AI Lineage Assistant, which reduces what used to take several days of manual analysis to five to ten minutes, with 10x to 100x acceleration across broader governance workflows. In the session “Using Active Metadata to Support Data Agents,” Gartner analyst Mark Beyer presented research showing that metadata volume grows exponentially with agentic AI.^[4] Manual approaches to lineage and governance won’t survive that scale. Organizations like LSEG that automated their metadata workflows early have a compounding advantage over those still relying on spreadsheets and tribal knowledge.

AI trust starts with what you can see

Every conversation I had at the summit circled back to trust. De Simoni found that more than 50% of vendors identify trust as the top barrier to agentic AI adoption.^[5] Gartner expects unsupervised AI deployment to remain below 10% through 2028. The industry is building AI agents faster than it’s building the trust infrastructure to support them.

Philip put it simply: “If we can’t see it, if we can’t understand it, how do we trust it?” Solidatus renders data lineage as interactive visual maps rather than rows of metadata in a spreadsheet. Visualization isn’t a nice-to-have for governance. When people can see their data lineage mapped out and confirm it matches their understanding of the organization, they trust it. When they’re poring over raw metadata for hours, they generate questions, not confidence.

That principle extends to AI outputs as well. Solidatus built hallucination protection directly into the AI Lineage Assistant. If the LLM returns a response that isn’t grounded in metadata within the platform, the system rejects it and forces a new attempt. The response has to be anchored in real data before it reaches the user. In financial services and other regulated industries, where human-in-the-loop oversight is standard, that validation layer is non-negotiable.

Terrence described how trust and lineage connect at enterprise scale. LSEG’s data trust program is built on four elements of trust, with Solidatus providing the lineage foundation.

“If we don’t understand what that data is, it’s very difficult for us to understand how we can use it, how we should use it, what value it can provide,” Terrence said.

Trust becomes even more critical as AI agents grow more autonomous. Philip pointed out that much of what vendors call “AI agents” today are chatbots running on request-response. True agentic AI creates its own plan, executes across 20 to 50 steps, and self-corrects along the way. Without lineage and metadata infrastructure, organizations have no way to verify what an agent did or why.

What to do with the infrastructure you already have

The data lineage and metadata systems that compliance teams built over the past decade are becoming the critical infrastructure layer for AI trust, AI agents, and AI governance. LSEG proved that by turning their lineage program into a strategic asset and a commercial data product. Solidatus proved it by extending a governance platform into an AI-accelerated workflow engine.

If your organization has invested in data lineage for compliance, the next step isn’t building a separate AI governance program. Audit what you already have. Identify where it covers AI use cases. Close the gaps. If you lead an AI or data science team, ask your governance counterpart what lineage coverage already exists for your training data, production models, and agent workflows. The organizations that connect these functions now will govern AI with confidence. The ones that start from scratch will spend the next two years catching up.

Listen to the full conversations with Philip Dutton, Terrence Hedin, and Caleb Watkins on the Data Faces Podcast.

Based on insights from Philip Dutton, CEO and Founder at Solidatus, Terrence Hedin, Data and Metadata Platform Director at LSEG, and Caleb Watkins, Solutions Engineer at Solidatus, featured on the Data Faces Podcast.

Frequently asked questions

What is Solidatus?

Solidatus is a data lineage and metadata management platform that maps, visualizes, and governs data flows across the enterprise. It is used primarily by financial services and other regulated industries to track how data moves through systems, meet compliance obligations, and build organizational trust in data. The platform recently introduced an AI Lineage Assistant that adds natural language interaction, automated code scanning, and regulatory compliance assessment to its existing governance capabilities.

Why is data lineage important for AI governance?

Data lineage documents where data comes from, how it moves through systems, and what obligations are attached to it. Those obligations, including purpose limitations, storage rules, and sharing boundaries, apply to AI the same way they apply to BI dashboards or regulatory reports. Organizations with mature lineage programs can extend their existing governance operating model to cover AI use cases without building a separate framework. Gartner research presented at the 2026 D&A Summit showed that organizations with graduated trust models achieve 64% compliance success compared to 23% without them.

How does data lineage differ from AI governance?

Data lineage is a component of AI governance, not a separate discipline. Lineage tracks how data flows through an organization and what happens to it along the way. AI governance addresses the broader question of how to ensure AI systems use that data responsibly. The argument from practitioners at LSEG and Solidatus is that the lineage and metadata infrastructure built for regulatory compliance already provides the semantic foundation AI agents need. Rather than creating a parallel AI governance program, organizations can extend what they have.

What is a bring-your-own-LLM model for data governance?

A bring-your-own-LLM model allows organizations to connect their own large language model to a governance platform rather than sending data through a vendor’s AI infrastructure. Unlike vendor-hosted AI models that route customer data through external systems, the BYOLLM approach keeps all data processing within the customer’s own environment. Solidatus uses this approach for its AI Lineage Assistant, meaning no data flows through Solidatus or any third party. This design addresses the primary security concern enterprises have about AI in governance contexts, particularly in regulated industries like financial services.

How does Solidatus prevent AI hallucinations in governance workflows?

Solidatus built hallucination protection directly into the AI Lineage Assistant. When the LLM generates a response, the system validates it against metadata that exists within the platform. If the response isn’t grounded in real data, the system rejects it and forces a new attempt. The response has to be anchored in verified metadata before it reaches the user. This approach ensures that AI outputs in governance contexts are based on actual organizational data rather than fabricated information.

Where should organizations start with AI governance if they already have data lineage?

Start by auditing your existing lineage coverage to identify where it already applies to AI use cases. Philip Dutton, CEO of Solidatus, argues that organizations don’t need a new operating model for AI governance because the one they already use for compliance works. LSEG provides a proof point, having evolved their lineage program from a regulatory tool into a strategic asset and commercial data product. The key is closing gaps rather than starting from scratch.

Podcast highlights

Philip Dutton, CEO and Founder, Solidatus (~15 min)

[0:00] Introduction at the Gartner D&A Summit

[0:28] What is Solidatus and why data lineage matters

[0:54] Data lineage meets AI governance

[1:50] The AI Lineage Assistant and natural language interaction

[3:17] Trust in AI and trust through lineage

[4:45] Human in the loop for financial services

[5:14] Why visualization builds data trust

[7:05] You can’t automate what you don’t understand

[8:27] Data lineage as AI lineage, same operating model

[9:29] What’s on attendees’ minds at Gartner

[10:48] True agentic AI vs. chatbots

[12:00] The future of Solidatus and agentic orchestration

[14:18] LSEG session preview and closing

Terrence Hedin, Data and Metadata Platform Director, LSEG (~6 min)

[0:00] Introduction and upcoming LSEG session preview

[0:23] Overview of the LSEG talk with Philip Dutton

[2:26] Lineage as a first class citizen in the age of AI

[3:32] From regulatory reporting to strategic asset

[4:47] How the Solidatus AI Lineage Assistant is changing workflows

[5:54] Session details and closing

Caleb Watkins, Solutions Engineer, Solidatus (~4 min)

[0:00] Introduction at the Gartner D&A Summit

[0:22] The AI Lineage Assistant and bring-your-own-LLM

[0:44] Trust and security in the AI agent

[1:01] Use case: AI-powered code scanning

[1:42] Days to minutes with automated lineage

[2:01] Use case: regulatory compliance (BCBS 239, AI Act)

[2:56] Customer feedback on the assistant

[3:29] Find Solidatus at Booth #929

About David Sweenor

David Sweenor is the founder and host of the Data Faces podcast, where he talks with the people who are making data, analytics, AI, and marketing work in the real world. He is also the founder of TinyTechGuides and a recognized top 25 analytics thought leader and international speaker who specializes in practical business applications of artificial intelligence and advanced analytics.

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence. David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn

^[1] Ronthal, Adam, and Georgia O’Callaghan. “Opening Keynote: The State of Data and Analytics.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[2] De Simoni, Guido. “Trust as the New Currency.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[3] García-Rodeja, Andrés. “How to Build the Context Layer for Reliable AI Agents.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[4] Beyer, Mark. “Using Active Metadata to Support Data Agents.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[5] De Simoni, Guido. “Trust as the New Currency.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

When Michael Meyer told his high school guidance counselor he was skipping university to attend a trade school for programming, the response was blunt: “All the computer programs that have ever needed to be written have already been written. You’re going down the wrong path.”

That was the late 1980s. Michael ignored the advice, enrolled anyway, and spent the next 35 years building a career across programming, data architecture, product marketing, and solutions engineering. He’s currently a Solutions Engineer at Snowflake, helping enterprise customers in Omaha get more out of their data platforms.

Today, a new generation of data professionals is hearing a familiar refrain. AI will write your code. AI will build your dashboards. AI will make your job obsolete. I recently sat down with Michael on the Data Faces podcast, and his career offers a compelling counterargument. The skill that carried him through every industry shift was something most people in data overlook entirely. Storytelling.

“The writing about the craft beer is more about the people, the historical significance of places and things like that, than it is about the beer.” — Michael Meyer, Solutions Engineer, Snowflake

About Michael Meyer

Michael Meyer is a Solutions Engineer at Snowflake, where he supports enterprise customers across the Omaha, Nebraska market. His career spans over 35 years in programming, data architecture, data governance, and product marketing at companies including Alation. He is also the author of Joe’s Brew Reviews, a book about Nebraska’s craft beer scene that’s really about the people and stories behind the breweries. In our conversation on the Data Faces Podcast, we discuss:

• Why storytelling is the most durable skill in a data career

• How the semantic layer went from a BI footnote to AI’s missing piece

• What vibe coding gets right and where it falls short

• The fundamentals that early career data professionals need to focus on

A storyteller who speaks data

Michael’s relationship with storytelling started at home. His father was a logical thinker who passed down an analytical mindset, while his maternal grandfather was the kind of natural storyteller you’d sit and listen to for hours. Michael found himself torn between journalism and programming after high school. He chose programming, but the storytelling instinct never went away.

On one side of the family, he got the analytical instinct, and on the other, the urge to make people lean in and listen. That combination connects every role he’s held, from programming to data architecture to product marketing to solutions engineering. Each demanded a different technical skill set, but each rewarded the ability to make complex things understandable. Storytelling wasn’t a hobby Michael kept separate from his day job. It became the thing that made every job better.

Walt the data janitor and the power of internal marketing

When Michael transitioned into data governance at a financial services company, the team needed people across the organization to care about data quality, catalogs, and standards. Instead of frameworks and compliance language, they created a fictional character named Walt.

Walt showed up across the entire data management program. When the team discussed data quality, Walt was the janitor pushing a broom through messy datasets. When they introduced data architecture concepts, Walt played a different role. He gave people a relatable way into abstract subject matter, and it worked.

“We built up our data program from creating a fictional character that helped us show each step of the way, from data architecture to quality. His name was Walt. When we talked about data quality, Walt was the janitor pushing the broom through.” — Michael Meyer, Solutions Engineer, Snowflake

That experience crystallized something for Michael. Getting people to care about data is a communication challenge, not a technical one. His internal marketing work caught the attention of Alation’s field marketing team, and they recruited him into a technical product marketing manager role.

The mindset shock of marketing

Moving into product marketing forced a shift in how Michael communicated. The real breakthrough came from learning to listen for the exact phrases customers used to describe their problems and then putting those words on the page rather than his own.

“There would be key phrases that would come out that customers would say, and if you could use those, especially within what you’re trying to portray, that’s where you could get the hook. You’d better get them interested right away. How do you get to that emotional side of somebody so that they think, ‘Wait a minute, that’s me’?” — Michael Meyer, Solutions Engineer, Snowflake

That customer-language instinct turned out to be portable. Michael is now back on the technical side as a Solutions Engineer at Snowflake, where he helps key enterprise accounts in the Omaha market. His marketing background makes him better at translating a complex platform into terms that resonate with the people who actually use it. Marketing gave him a vocabulary for the rest of his data career.

The semantic layer is a storytelling problem

If you’ve worked in data long enough, you remember defining semantic layers inside BI tools like Cognos Framework Manager, giving human-readable names to cryptic database columns so business users could build their own reports. Nobody called it storytelling at the time, but that’s exactly what it was.

The concept never went away, but it stayed fragmented. Every BI platform maintained its own semantic definitions, which meant every tool told a slightly different version of the truth. AI changed the equation. When a business user asks a natural language question of an AI assistant, that assistant needs to understand the question in the terms the business actually uses, not technical column names or another company’s jargon, but the specific vocabulary of that organization.^[1]

A modern semantic layer provides that context by defining facts, dimensions, metrics, relationships, and business rules in a format that both humans and large language models can interpret. Without it, AI tools produce answers that sound plausible but miss the mark, and once business users lose trust, they rarely come back.

“If I’m going to talk with my data, I need to talk to it in terms of how the business speaks with the data, not technical terms, not how another financial company talks to theirs, but how my organization talks. That’s really the key.” — Michael Meyer, Solutions Engineer, Snowflake

Gartner predicts that organizations will abandon 60% of AI projects unsupported by AI-ready data through 2026.^[2] Michael estimates that building a good semantic model today is still about 70% human work. AI can generate descriptions and suggest metrics, but it can’t replace the institutional knowledge of someone who has spent years working with the data. A subject matter expert knows why two fields that look similar mean very different things in practice.

AI-assisted coding and the proof-of-concept trap

“Vibe coding” is the practice of using AI to generate code from natural language prompts rather than writing it by hand. Michael has been experimenting with Snowflake’s Cortex Code and recently used it to build a machine learning pipeline from scratch. He fed it a retail dataset, picked a use case focused on detecting late delivery issues, and let it design the full pipeline in a notebook.

The first model returned about 50% accuracy. Michael knew that was unacceptable. He iterated with Cortex Code, adding feature engineering and additional training, until the model reached 85% accuracy. The critical skill in that process came down to judgment, knowing that 50% meant the model was broken and that 85% meant it was worth showing to a data scientist for validation.

AI-assisted coding compresses timelines and lowers the barrier to exploring new technical domains. But it’s not a substitute for the judgment that comes from understanding your data. Someone who can’t read the story the numbers are telling, who can’t look at a metric and know whether it makes sense in context, will eventually ship something dangerous.^[3]

“If you don’t have any understanding of how to test and verify, and if you’re just taking everything AI does as being 100% accurate, that could quickly actually become a career ender.” — Michael Meyer, Solutions Engineer, Snowflake

What endures when everything else changes

Michael’s guidance counselor was wrong about programming in the late 1980s. The people saying AI will replace data professionals are making the same mistake today. The tools and platforms will keep changing, but the fundamentals don’t.

When I asked Michael what early career data professionals should focus on, he didn’t start with AI prompting or the latest framework. He started with data modeling. Understand what good data looks like. Learn how raw data becomes something a business user can actually consume.^[4] Know how to test and validate results, because if you can’t verify what AI gives you, you have no business putting it in front of a decision-maker.

Michael pointed out that he’s met AI engineers who have never worked with data before, and that gap shows up when models need to connect to real business outcomes. As Thomas Davenport and Randy Bean argue in MIT Sloan Management Review, 2026 is the year the industry must shift from chasing AI hype to realizing actual enterprise value, and that shift depends on exactly the kind of foundational data skills Michael is describing.^[5]

“Find something that energizes you, what are some of your strengths, and lead into those. It took me a long time to really make my writing part of my career. But it can be done.” — Michael Meyer, Solutions Engineer, Snowflake

Beyond the technical fundamentals, Michael’s advice is to get out from behind the screen. He runs meetups in Omaha and credits the networking he’s done over the past decade with keeping him grounded and energized. Be a constant learner, he says, but learn from people, not just platforms.

Not all programs have been written. And the stories haven’t all been told.

Listen to the full conversation with Michael Meyer on the Data Faces Podcast.

Based on insights from Michael Meyer, Solutions Engineer at Snowflake, featured on the Data Faces Podcast.

Podcast highlights

[0:00] Opening and introduction

[2:00] Michael’s 35-year career across programming, data architecture, and Snowflake

[4:00] Joe’s Brew Reviews, journalism vs. programming, and the storytelling instinct

[6:30] Walt the data janitor and how a fictional character made data governance relatable

[11:00] The mindset shock of moving from data architecture to product marketing at Alation

[14:00] Customer language, emotional hooks, and storytelling on a B2B web page

[17:00] Coming back to the technical side as a Solutions Engineer at Snowflake

[19:00] What the semantic layer is and why AI made it urgent

[23:00] Facts, dimensions, metrics, verified queries, and business rules in a semantic model

[25:30] Building a semantic model: 70% human work and why institutional knowledge matters

[28:30] Vibe coding with Snowflake Cortex Code and iterating from 50% to 85% accuracy

[32:00] Why early career data professionals should start with data modeling fundamentals

[34:30] Find what energizes you, get out from behind the screen, and be a constant learner

[35:30] Craft beer recommendations and closing

About David Sweenor

David Sweenor is the founder and host of the Data Faces podcast, where he talks with the people who are making data, analytics, AI, and marketing work in the real world. He is also the founder of TinyTechGuides and a recognized top 25 analytics thought leader and international speaker who specializes in practical business applications of artificial intelligence and advanced analytics.

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence. David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn

^[1] David Sweenor, “Your AI Doesn’t Have a Model Problem. It Has a Data Context Problem,” TinyTechGuides, February 24, 2026, hcansubject-matter ttps://tinytechguides.com/blog/your-ai-doesnt-have-a-model-problem-it-has-a-data-context-problem/

^[2] Gartner, “Lack of AI-Ready Data Puts AI Projects at Risk,” Gartner Newsroom, February 26, 2025, https://www.gartner.com/en/newsroom/press-releases/2025-02-26-lack-of-ai-ready-data-puts-ai-projects-at-risk

^[3] David Sweenor, “Data Lineage for AI: Why Truth Beats Hope in Banking,” TinyTechGuides, December 2, 2025, https://tinytechguides.com/blog/data-lineage-for-ai-why-truth-beats-hope-in-banking/

^[4] David Sweenor, “From ‘AI-Ready’ to AI Reality: Why Actionable Data Strategies Beat Endless Planning,” TinyTechGuides, June 3, 2025, https://tinytechguides.com/blog/from-ai-ready-to-ai-reality-shane-murray-on-data-trust-and-why-action-beats-planning/

^[5] Thomas H. Davenport and Randy Bean, “Five Trends in AI and Data Science for 2026,” MIT Sloan Management Review, 2026, https://sloanreview.mit.edu/article/five-trends-in-ai-and-data-science-for-2026/

I’ve often equated people with birds – they’re always chasing the next shiny object that comes along, in this case, agentic AI. Sadly, I’ve watched a scene like this play out at nearly every company I’ve worked with. A team builds an AI assistant, loads it with pristine data, asks it well-crafted questions, and the demo goes flawlessly. Leadership sees the result and greenlights production.

Six months later, the project has lost its luster, the team is frustrated, and the models underlying the assistant that looked brilliant in a conference room can’t survive contact with the real world. What happened? The model didn’t break. The environment changed, and the carefully curated conditions that keynote the shiny demo simply don’t exist at scale across a messy, fragmented enterprise.

In a recent episode of the Data Faces Podcast, I sat down with Asa Whillock to talk about what it actually takes to move AI from pilot to production. Asa has spent 35 years in software and has lived this challenge from every angle, inside large enterprises like Adobe and Alteryx, and now as the founder of his own AI company. His core argument is one that every data and AI leader needs to hear: the distance between a brilliant demo and a production system has very little to do with model capability. It’s a data context problem. And most organizations are only scratching the surface of what “context” really means.

“When you think about what makes AI production-ready, it is really not so much about the model. When you talk about demos and pilots, you’re almost always talking about a cultivated set of data and a cultivated set of questions so that the result is just outstanding. You’re like, this is amazing, why would we not deploy this everywhere?” — Asa Whillock, CEO and Founder, Euphonic AI

About Asa Whillock

Asa Whillock is the CEO and founder of Euphonic AI, a growth acceleration agent serving revenue operations and demand generation leaders who engineer organizational growth. His career spans 35 years in software across major enterprises, including Adobe, Alteryx, Intel, and AOL, with deep experience in analytics, product, and go-to-market strategy. He also spent two years performing stand-up comedy, which explains his talent for making complex enterprise AI concepts feel accessible through unexpected analogies. In our conversation on Episode 32 of the Data Faces Podcast, we discuss:

• Why AI pilots deceive leadership about production readiness

• The three categories of data context that most organizations are missing

• Why chasing frontier models is a distraction from the real work

• How to connect AI investment to the metrics that actually drive your business

The three buckets of context that make AI production-ready

So if the model isn’t the problem, what is? Asa frames it with an analogy that will feel familiar to anyone who has worked inside a large organization: enterprise data is “scattered like toys across a two-year-old’s bedroom.” From the outside, you’d assume that companies with thousands of employees and massive technology budgets have their data neatly organized and ready to fuel AI systems. Well, you know what they say about assumptions, and nothing could be further from the truth. It’s the reason so many promising AI initiatives have failure to launch on their way to production.

Asa breaks the challenge into three categories of context that AI needs to function at production quality. Most organizations are actively working on the first category and are barely aware that the other two exist.

Machine-driven data context

This is the layer that gets the most attention. Machine-driven data is the row-level, tabular data living in your CRM, ERP, and HRM systems. Data teams have been working with this information for years, and the challenge here isn’t a lack of awareness. The challenge is that it’s spread across roughly 350 systems per enterprise, and each system has its own opinion on what data should look like. Connecting them is painful, expensive, and slow, but at least organizations recognize this work needs to happen.

Operational metadata context

This is the layer most teams overlook entirely. Operational metadata includes the configuration settings, workflow routing rules, and log files that determine how data actually moves through an organization. Asa describes this as “the train control that determines if this train goes to Seattle or Albuquerque.” A single configuration switch can change the entire path a customer record follows, and that switch often lives buried in a UI screen that no API can reach. When AI operates without awareness of these controls, it’s making recommendations based on data flows it doesn’t actually understand.

Human decision data context

“Think about the institutional memory of your organization. You can almost always name that person. Imagine living your life without that person in every decision you ever made. You’d be this uninformed AI, guessing. When you have the machine data, the metadata, and the human decision context together, now your AI is ready for production.” — Asa Whillock, CEO and Founder, Euphonic AI

This is the hardest layer to capture and the most valuable once you have it. Human decision data is the institutional knowledge that lives in people’s heads, the reasoning behind why an organization made specific choices, abandoned certain approaches, or accepted particular architectural tradeoffs. Every organization has an institutional memory holder who knows where the skeletons are, the person everyone turns to when they need to understand why a decision was made three years ago. That person knows which approaches were tried and abandoned, which compromises became permanent, and which workarounds nobody ever documented. Asa puts the stakes plainly: imagine making every decision in your organization without that person’s knowledge. That’s the position your AI is operating from right now if you haven’t captured this layer.

The context gap is a $130 billion problem

Each of these gaps is costly on its own, but the compound effect of missing two or three layers simultaneously is what stalls most AI initiatives. Kimberly at Andreessen Horowitz has pointed out that 9 out of 10 automations that could exist today simply don’t, because unlocking the data needed to build them is too difficult. By her firm’s estimate, that represents a $130 billion opportunity in context gaps that have nothing to do with model capability.

Stop chasing models and start hydrating with data context

The industry’s obsession with frontier models is understandable. The capabilities are genuinely impressive, and as Asa puts it, “models are so hot right now.” For data science leaders, the temptation to evaluate, compare, and chase the latest release feels like productive work because the improvements are measurable and exciting.

But Asa argues that this focus is a distraction from the work that will actually determine whether your AI investments succeed or fail. He frames the choice in terms that are hard to ignore. Given the option between a frontier model paired with a poor data context and a slightly less capable model that has been fully hydrated with all three layers of context, the hydrated model wins every time. A world-class model operating without awareness of your operational metadata and institutional decision history will produce outputs that sound polished but miss how your organization actually works. A less flashy model armed with deep context will deliver results that make sense to the people who have to act on them.

“If you’re going to choose a model that is just at the absolute frontier of capability with poor data, or take a model that is maybe half a step backwards in capability but give it all of the context it needs to make an amazing outcome, I’ll tell you which one I will choose every time. Invest in breaking down those barriers of permissions, access, data, those unsexy things.” — Asa Whillock, CEO and Founder, Euphonic AI

The highest-leverage work available right now isn’t evaluating the next model release. It’s the unglamorous, difficult work of breaking down permission barriers, building cross-system data access, and finding ways to capture the institutional knowledge that currently lives only in people’s heads. That’s where production-ready AI gets built, and it’s the work most organizations keep deferring because it doesn’t generate the same excitement as a new model announcement.

Aim your context at the metrics that actually drive your business

Once you’ve mapped your three layers of context, you need to point them to a specific location. This is where Asa’s advice shifts from framework to action, and it starts with a question he hears constantly from leadership: if AI is so transformative, where is the ROI? Asa’s response is to turn the question back on the person asking it.

“I ask leadership, what are the five things that drive your ROI? Have you deployed an AI solution to tune for that? Have you looked at what actually impacts that metric? Almost every time the answer is, well, no, I haven’t really dug into that. If you ask yourself those questions and drive into those key metrics, you will find that transformative ROI.” — Asa Whillock, CEO and Founder, Euphonic AI

What are the five or six metrics that actually drive your business? Not the vanity metrics on a dashboard, but the numbers that directly influence customer acquisition cost, speed to lead, time to value, or whatever levers matter most for your organization. And once you’ve named them, have you dug into what drives those metrics? Most of the leaders Asa talks to haven’t done this work. They’ve deployed AI against surface-level outcomes and are puzzled when the results feel incremental rather than transformational.

One of the most formative experiences in Asa’s career was watching Adobe build what they called a data-driven operating model. Adobe didn’t just track top-line metrics like customer adoption and retention. They decomposed each metric three to five layers deep to understand what specifically influenced it, and then what influenced those influencing factors. That level of decomposition is what turned their AI investments from interesting experiments into systems that moved the business forward.

Next steps

“You can’t be standing back, going, where’s the ROI? You have to have visibility. You have to have headlights on how the metrics that really drive your business, and they’re really tuning up. Focus on what matters for you.” — Asa Whillock, CEO and Founder, Euphonic AI

The organizations that win with AI over the next few years won’t be the ones running the most capable models. They’ll be the ones who did the unglamorous work of stitching their data context together across systems and connecting it to the metrics that actually matter for their business.

If you’re a data or AI leader trying to figure out where to focus, Asa’s framework gives you a practical starting point.

• Audit your three buckets of context. For your highest-priority AI initiative, map which layers you actually have coverage on today. Most teams will find they’re reasonably strong on machine-driven data and nearly empty on operational metadata and human decision context.

• Decompose the metrics that matter for your business. Name the five or six numbers that actually drive your outcomes and dig three to five layers deep into what influences them. That decomposition is where you’ll find the specific places to aim your AI investments.

• Capture institutional knowledge before it disappears. Identify the three to five people in your organization who hold the decision-making context that no system has ever recorded, and start documenting what they know.

Listen to the full conversation with Asa Whillock on the Data Faces Podcast.

Based on insights from Asa Whillock, CEO and Founder at Euphonic AI, featured on the Data Faces Podcast.

Share

Leave a comment

Podcast Highlights - Key Takeaways from the Conversation

Podcast highlights

[0:51] Asa introduces his 35-year career in software and the founding of Euphonic AI

[2:22] The Voltron analogy, stand-up comedy, and making enterprise AI concepts accessible

[4:16] Why large enterprises are “relentlessly vertical” and how that creates friction for AI

[7:28] The ebb and flow between horizontal and vertical software cycles

[8:39] Why AI pilots deceive you: cultivated data versus the reality of production

[10:00] The three buckets of data context: machine-driven data, operational metadata, and human decision data

[13:17] The role of unstructured data and why operational context matters more than document archives

[15:01] “What do you want to be great at?” and why companies shouldn’t pivot to becoming AI companies

[19:16] Vibe coding, competitive parity, and why adding the same capabilities as everyone else nets to nothing

[22:24] How to align AI investments with your business differentiation instead of chasing technology

[25:51] Why data context matters more than model capability and the case for the “half-step-back” model

[29:25] The bridge between systems of record and why nobody is incentivized to build it

[32:29] Asa’s one piece of advice: identify the five metrics that drive your business and dig three to five layers deep

About David Sweenor

David Sweenor is the founder and host of the Data Faces podcast, where he talks with the people who are making data, analytics, AI, and marketing work in the real world. He is also the founder of TinyTechGuides and a recognized top 25 analytics thought leader and international speaker who specializes in practical business applications of artificial intelligence and advanced analytics.

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence. David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

Follow David on Twitter@DavidSweenor and connect with him on LinkedIn.

Gene Arnold has never been more excited about a technology in his career, which is exactly why he’s become an active voice for pragmatic governance. Gene works as a partner sales engineer at Atlan, where he came to AI through the data catalog world and discovered that models living on top of data made governance unavoidable.

But his perspective extends beyond enterprise software. He runs a GitHub AI agent engineering course, builds with 3D printers and stepper motors, creates his own semantic generators, and hosts a YouTube channel on mountain biking that taught him how social media lets practitioners break into industries without turning pro. Gene sees governance as the foundation that lets good ideas reach production.

About the guest

Gene Arnold is a partner sales engineer at Atlan, where he works closely with partners like Snowflake and Databricks to help organizations navigate governance, modern data stacks, and AI use cases. In this conversation, we discuss:

• Why the people most excited about AI often care most about governance

• How pressure from leadership and individual contributors creates gaps

• Why models persist bias rather than create it

• The role metadata and semantic layers play in AI accuracy

• How to pick your first AI project and what to watch for

The dual pressure problem

When Gene talks with organizations about AI adoption, he sees the same recurring pattern. Pressure to move fast comes from leadership and individual contributors alike, and most companies have no framework for managing what gets built.

From the top, leadership sees competitors moving and asks why the organization isn’t keeping pace. AI isn’t a hard concept to grasp when everyone’s talking to ChatGPT. Executives don’t need technical depth to feel the strategic urgency. As Gene put it, “Top down, I don’t have to be a rocket scientist to understand we need it.”

From the bottom, individual contributors can now build AI tools without knowing how to code. Gene pointed to how quickly someone can knock out a workflow with tools like n8n, with templates available for almost everything. “Bottom up, I don’t have to be a rocket scientist to build something.”

That collision is where governance breaks down. Gene described the pattern he sees repeatedly: “Look what I made. Look what I made. Okay, hey, whoa, they made it. Send it out. Whoa, stop.” Roughly 80% of AI projects never reach production. Organizations are building plenty, but the challenge is building things that survive scrutiny.

“We don’t want to stop innovation. That’s a bad thing. But here’s the box. Let’s try to stay somewhere in the box.” — Gene Arnold, Partner Sales Engineer, Atlan

The box is a container for evaluating ideas before they hit production. Two accountability questions expose most governance gaps before they become public failures. First, do we have the right to do this? Gene cited an example of an organization that used facial recognition without proper authority to include faces in its models. The technology worked fine, but the authorization didn’t exist, and legal exposure followed. Second, what did we train it on? Traceability and lineage matter because models inherit whatever lives in the training data, including the biases and foibles that data contains.

Organizations that skip governance early often end up locking down AI entirely after a public failure. A little structure now prevents overcorrection later.

Models don’t create bias, they persist it

Even organizations that answer both accountability questions correctly at launch can find themselves exposed months later. Governance requires ongoing attention beyond the initial launch.

“A model doesn’t create bias. A model simply persists bias. The model was trained on X. If X was biased, then the model is biased.” — Gene Arnold, Partner Sales Engineer, Atlan

By now, the case of resume screening models that have passed over qualified female candidates is well-documented. Models trained on historical hiring data learned to predict what companies had historically done, including decades of biased evaluation. The teams involved may have done proper diligence, and the model performed exactly as designed. That was the problem.

Gene’s takeaway is direct: “I can’t train a model on the future. I don’t know what the future holds. I have no data on the future.” History carries black marks, and models trained on history inherit them.

The challenge gets harder when one variable masks another. Gene described research he conducted on AI bias using loan approval data. A model might show that PhD holders get approved at higher rates than those with bachelor’s degrees, which seems reasonable until you remove the PhD variable, and a different pattern emerges. The PhD status was masking discrimination based on other factors, like zip code. As Gene explained, “Sometimes one feature in the model can outweigh something else, and all of a sudden, when that one’s turned off... the PhD covered the fact that down here, loan payment wasn’t always paid in full.”

The answer is continuous testing. Gene recommended running tests against thousands of synthetic records, varying demographic features to see if outcomes shift unexpectedly. The goal isn’t perfection. “Constant testing, tweaking the levers to make sure that you properly, at least make it as unbiased as you can.”

Humans also need to remain part of the decision chain. You “can’t just flip the switch and say, best of luck, and let this thing go all day long and hope that you get it right. That’s scary.”

Governance starts with knowing what you have

Bias isn’t the only blind spot. Many organizations don’t know what data they have, which makes governance impossible before the model is even built.

Gene sees this play out in a simple scenario. Ask a model for “a summary of East Coast Q1 sales.” The model doesn’t know what Q1 means to your organization. Does your fiscal year start in January? What territories count as the East Coast? Without semantic information, the model guesses based on patterns in historical queries or fails entirely.

“Metadata is literally the deciding factor on how a model answers correctly.” — Gene Arnold, Partner Sales Engineer, Atlan

“If you don’t give it that semantic knowledge, it doesn’t know how to answer some of these questions,” Gene explained. This extends to any company-specific metric. How does the organization calculate customer lifetime value? What counts as an active user? The model needs that context, and the context needs to be governed.

The problem compounds when organizations have multiple versions of the same information floating around. Gene described scenarios where 10 versions of the same manual exist across different systems. Which one should train the model? Without proper curation through a data catalog or similar system, you’re building on uncertain foundations.

His practical workaround acknowledges reality: “What I can do now with AI is feed all 10 of those versions into a model, and now ask it questions... and it’ll respond well, based on the knowledge that I have, these are the three different methods that I feel would be appropriate, and that’s where the human judgment comes in.”

Start with what already works

Gene’s advice for getting started runs counter to the instinct to use AI to fix broken processes. “Pick a process that you know is working well, but you want to automate. Unless you know what good looks like, how do you know if the model is doing well?”

“Automation can either make things really good really fast, or make things really bad really fast.” — Gene Arnold, Partner Sales Engineer, Atlan

The reasoning is straightforward. If you don’t know what correct output looks like, you can’t evaluate whether the model is helping or hurting.

Gene distinguished between processes that just need speed and processes that need complete rethinking. “There are times where, well, you know what, we needed someone to read all these files. We still think they need to be read, but I’m going to have a model run them now. The process is fine, but the model can read it faster than a human.” Document summarization, internal search across knowledge bases, and data quality checks, which use predefined rules, all fit this pattern.

The opposite scenario requires more caution. “There are other times, like the Excel spreadsheet, where we’re going to say, well, now’s a good time to bring in a new system, right? And retool.” Customer-facing agents, automated decision-making on loans or hiring, and any process where the existing approach is broken are poor candidates for a first AI project.

Before any project goes live, leaders should be able to answer a handful of questions. Who owns this model? Who trained it and on what data? What does correct output look like? Do we have the rights to use this data in this way? What’s our plan for ongoing monitoring?

Gene noted that the first project teaches more than the use case itself. “You’re going to learn if your team works well together, is your AI governance workflow proper, right? So you’re going to learn a ton by just picking that one project that you already know how it’s supposed to end.”

One win builds trust, and trust enables the next project. Governance creates the conditions for that first win to happen safely.

Build something beautiful with it

Throughout our conversation, Gene kept returning to the same idea: governance exists to protect your ability to keep building. His philosophy on AI comes down to a simple metaphor.

“AI is just like any other tool. I can take a hammer and I can build something beautiful with it, or I could take that hammer, and I could smash something beautiful.” — Gene Arnold, Partner Sales Engineer, Atlan

For data science leaders, the challenge has shifted from whether to adopt AI to whether you can sustain it without creating problems that force you to backtrack later. The organizations that keep momentum will be the ones that built governance into the process from the start.

Gene’s parting point: “AI is not going away... this is the time where you’ve got the chance, because AI is still pretty new, to embrace it, understand what it can do and build something beautiful with it.”

Listen to the full conversation with Gene Arnold on the Data Faces Podcast.

Based on insights from Gene Arnold, Partner Sales Engineer at Atlan, featured on the Data Faces Podcast.

Share

Leave a comment

Podcast Highlights - Key Takeaways from the Conversation

Podcast highlights

[1:10] Gene’s path from DJ to AI advocate Gene explains how he accidentally fell into sales engineering after years as a DJ and discovered it was “a nerd’s stage to present and have fun.” His work at Atlan with data catalogs led him into AI, which he calls his “new love and passion.”

[7:05] How data governance and AI governance relate Data governance focuses on keeping data clean, accurate, and unbiased. AI governance focuses on decision-making and how systems behave. Gene explains how the two work together: “It’s an agent. Let’s govern what it’s allowed to do and how, and then when it does it, what data is it actually going to respond back?”

[9:56] Why unstructured data is the gold mine Gene argues that 80% of valuable information lives in emails, documents, and support conversations. “That’s where your answers are. That’s where your conversations are.” Smart companies are extracting insights from customer support chat logs to understand what products are working and what’s failing.

[12:28] The duplicate data problem Most organizations have multiple versions of the same information scattered across systems. Gene’s practical workaround: feed all 10 versions into a model and let it surface the best approaches, then bring in human judgment to make the final call.

[14:34] Why 80% of AI projects never reach production “It’s just so easy to make something. But then when you really put it down into the real world and run it, did you really properly QA this new cool thing?” Gene argues that companies are failing because they lack governance workflows to evaluate what’s being built.

[17:10] The dual pressure problem AI pressure comes from leadership and individual contributors simultaneously. “Top down, I don’t have to be a rocket scientist to understand we need it. Bottom up, I don’t have to be a rocket scientist to build something.” This creates a collision that governance needs to manage.

[19:18] The resume screening bias example Gene walks through the well-known case of AI models penalizing feminine-coded language in resumes. “A model doesn’t create bias. A model simply persists bias. The model was trained on X. If X was biased, then the model is biased.”

[21:40] The two accountability questions Gene identifies the questions that expose governance gaps: Who designed the model? What did you train it on? He also cites an example where an organization used facial recognition without proper authority to use the faces in their models.

[23:56] How one variable can mask another Gene describes his research on AI bias using loan approval data. PhD holders got approved at higher rates, but when you removed that variable, discrimination based on zip code emerged. “Sometimes one feature in the model can outweigh something else.”

[27:02] Start with a process you know is working “Pick a process that you know is working well but you want to automate. Unless you know what good looks like, how do you know if the model is doing well?” Gene advises against using AI to fix broken processes as a first project.

[28:15] Automation amplifies everything “Automation can either make things really good really fast, or make things really bad really fast.” Gene explains that your first project teaches you more than just the use case: you learn if your team works well together and if your governance workflow is functioning.

[31:29] Why metadata is the deciding factor Gene introduces the semantic layer concept: models need context to answer questions correctly. Without knowing what “Q1” means to your organization, a model will guess or fail entirely. “Metadata is literally the deciding factor on how a model answers correctly.”

[36:10] Gene’s parting advice “AI is just like any other tool. I can take a hammer and I can build something beautiful with it, or I could take that hammer and smash something that was beautiful.” Gene encourages listeners to embrace AI while understanding its weaknesses so it gets used correctly.

About David Sweenor

David Sweenor is an expert in AI, generative AI, and product marketing. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter@DavidSweenor and connect with him onLinkedIn.

Randy Bean has spent four and a half decades watching organizations struggle with data. The tools keep getting better, faster, and cheaper. Yet companies keep failing for the same reasons they failed in 1988 when Randy was working on natural language processing (NLP).

The failures result from barriers that are both cultural and structural. Leaders remain reluctant to ask uncomfortable questions about organizational readiness, and C-suites have fragmented into competing data roles with overlapping mandates.

In his 2026 AI and Data Leadership Executive Benchmark Survey, Randy found that 94 percent of Fortune 1000 executives cite culture and people as their principal barrier to AI adoption, while only 6 percent point to technology. This gap explains why so many AI projects never make it past the pilot stage.

About the speaker

Randy Bean is a senior advisor and board member who has spent 4.5 decades working with executives at Fortune 1000 companies on data, analytics, and AI leadership. He founded and leads the annual AI and Data Leadership Executive Benchmark Survey, now in its 15th year, which captures insights from over 100 Chief Data Officers, Chief AI Officers, and Chief Analytics Officers from the Fortune 1000. His work has been published in Forbes, Harvard Business Review, MIT Sloan Management Review, and the Wall Street Journal.

In this episode, we discuss

• Why 94 percent of AI challenges are cultural rather than technical

• The “readiness test” that predicts whether organizations will succeed or stall

• Why legacy companies should stop benchmarking against digital disruptors

• The case for unifying CDO, CAO, and CAIO roles

• Why the best Chief Data Officers aren’t data geeks

• The 5-question, 5 percent framework that kills “boil the ocean” projects

The 94 percent problem hiding in plain sight

Randy didn’t set out to become a data expert. He studied English, History, and Art History in school. But when he needed a job, employers wanted technical skills, so he trained as a COBOL and assembler programmer. His interest was never really in the programming itself, but in the stuff you moved around, the data.

That background gave Randy a different lens than most technologists. Where others frame AI adoption as an engineering problem, he keeps coming back to the human side. People never really like to change on their own volition, and organizations are made up of people, which compounds the issue.

The 2026 survey data confirms this. When asked to identify the principal challenge to data and AI adoption, 94 percent of Fortune 1000 executives pointed to culture and people. Only 6 percent said technology was the barrier. This ratio has held steady for years. The technology has transformed dramatically. The human barriers haven’t.

Cultural barriers persist because they’re difficult to measure. You can audit a tech stack and benchmark infrastructure against competitors. But how do you assess whether an organization is genuinely ready to change how it operates?

Randy has developed his own informal test. He described folding up his notebook ten minutes into specific client meetings, sitting through the rest, already knowing the engagement wouldn’t go anywhere.

“They weren’t ready. You could tell it. They said, ‘We’ve got everything all figured out.’”

— Randy Bean, Senior Advisor and Board Member

When leaders claim they have everything figured out, they signal they’re not open to the difficult conversations transformation requires. Better tooling won’t fix that.

The challenge for data and AI leaders is learning to recognize these signals early, before investing months in initiatives that were never going to succeed.

Get your house in order

During the pandemic, the Chief Digital Officer of the nation’s largest insurance company shared something striking with Randy. “We’ve done more to execute on our digital transformation strategy in the past six months than we did in the previous 20 years.” The company moved because it had to. Customers couldn’t meet face-to-face, and online channels became existential overnight.

Transformation at legacy companies rarely happens because of vision or strategy. It happens when the alternative becomes untenable.

Randy’s survey data reveals an important distinction. Ninety percent of Fortune 1000 companies are legacy organizations. Only 10 percent belong to the “move fast and break things” crowd.

“90% of the Fortune 1000 are legacy companies... Those 10%, they can pioneer new things, but for the other 90%, they don’t need to compete against that other 10%—they just need to compete against one another.”

— Randy Bean, Senior Advisor and Board Member

When a 150-year-old manufacturer compares its AI maturity to a Silicon Valley startup, it creates false pressure that leads to poorly planned initiatives and wasted resources. The bank that figures out AI-driven fraud detection doesn’t need to outpace Google. It needs to outpace the other banks.

Stop measuring your AI maturity against tech-native outliers. Your real competition is the traditional competitor who figures out cultural readiness before you do.

Competitive pressure isn’t the only problem. There’s a structural one too. The proliferation of data-related C-suite titles has created confusion at many organizations. Randy’s 2026 survey found that 39 percent of companies have appointed a Chief AI Officer in addition to their existing Chief Data Officer. This leads to competing functions, unclear accountability, and redundant mandates.

In December 2025, Randy co-authored an article in Harvard Business Review with Vipin Gopal and Tom Davenport, arguing for a unified Chief Data, Analytics, and AI Officer role. About 30 percent of readers pushed back with legitimate concerns about the differences between AI and data governance operating models.

“We were just trying to create some level of sanity and clarity in the C-suite, so that you didn’t have all these competing and redundant functions, but rather came up with a unified mindset around how you manage data, analytics, and AI.”

— Randy Bean, Senior Advisor and Board Member

The point isn’t that unification is the only answer. The fact is that fragmentation has costs. Someone needs to be accountable for how data, analytics, and AI work together.

Hire for business acumen, not technical pedigree

The Chief Data Officer at JP Morgan Chase sits on the bank’s 14-person operating committee and reports directly to Jamie Dimon. Her previous role wasn’t in data engineering or analytics architecture. She was the Global Chief Investment Officer. Her questions focus on the most complex business problems the bank needs to solve and how data and AI can address them.

Diana Schulthaus, Chief Data Officer at Colgate-Palmolive, offered another perspective at a panel Randy moderated. When asked how much time she spends on offensive versus defensive data activities, her answer surprised the room. “I spend 100% of my time on offense.” The audience applauded.

In 2020, only 55 percent of organizations reported focusing on offensive, growth-oriented data activities. By 2025, that number jumped to 86 percent. The role has evolved from regulatory compliance to business strategy.

“You don’t need a data architect or data engineer or data modeler to be the chief data officer. You need a business leader who understands how data is going to be used so the organization can be more effective.”

— Randy Bean, Senior Advisor and Board Member

Randy learned this lesson early in his career. As a young COBOL programmer, he watched his technical colleagues return from meetings with business users and complain that the business people were stupid and couldn’t articulate what they wanted.

One day, he pushed back.

“They’re the people that employ us. They’re the people that go out and get the customers. They’re the people that do the business. We wouldn’t even be employed if it wasn’t for these people. So maybe we should give them some credit and figure out how to speak their language.”

— Randy Bean, Senior Advisor and Board Member

Effective data leaders understand enough about technology to separate nonsense from reality, but they lead with business problems rather than technical architecture.

The 5-question, 5 percent framework

Early in Randy’s career, he worked at a major bank that was part of Bank of America. He noticed the organization captured enormous amounts of customer information. One day, he asked a manager what they did with all of it.

“The regulators make us hold on to it for six years,” the manager said, “and then we put it in the furnace.”

The data existed. The infrastructure to store it existed. But nobody had asked what questions the data could answer.

He’s watched too many companies since then pursue what he calls the “boil the ocean” approach, trying to perfect every piece of data across the enterprise before putting any of it to use.

“Understand what the most important business questions that you need to ask are—not like 1,000, but like 5 or 10... Not all data is created equal, and sometimes 5% of the data is all that it takes to answer 95% of the questions.”

— Randy Bean, Senior Advisor and Board Member

Start with the five to ten most important business questions your organization needs to answer. Then identify the key data elements required to answer those specific questions. Resist the urge to build a comprehensive data infrastructure before you’ve proven value on the questions that matter most.

The same discipline applies to AI adoption timelines, and Randy encourages leaders to resist FOMO and the pressure to match competitors’ announcement cycles.

“Forget about the FOMO, the fear of missing out. Step back a little and think about where we’re going as a business. Where do we need to be? What capabilities do we need to have? How can AI augment those capabilities?”

— Randy Bean, Senior Advisor and Board Member

In 2023, only 5 percent of organizations had AI in production at scale. By 2024, that jumped to 24 percent. By 2025, it reached 40 percent. Progress came from organizations that planned thoughtfully, not from those that rushed to keep up with headlines.

What matters is where you end up in three, five, or ten years. Not what you announce this quarter.

Leading from the middle

Not every reader holds a CDO title or sits in the C-suite. Many data and AI practitioners work within organizations where leadership hasn’t embraced these shifts. They see the 94 percent problem from the inside but lack the authority to mandate change.

The notebook test works at every level. When your leadership team says they have everything figured out, you’re facing a readiness gap. More analysis or better dashboards won’t change the minds of those who aren’t open to change.

The 5-question framework scales both up and down. You don’t need enterprise-wide buy-in to prove value. Find one business question your team can answer with existing data. Deliver insight that helps someone make a better decision. Build from there.

“It does help to have some level of underpinnings so that you can separate the nonsensical from the realities... You gotta know enough to call BS when someone’s giving you a pitch.”

— Randy Bean, Senior Advisor and Board Member

Progress in AI adoption hasn’t come from organizations that moved fastest. It came from those who built readiness, clarified accountability, and focused on questions that mattered. That work can start anywhere in the org chart.

Call to action

Randy Bean’s full 2025 AI and Data Leadership Executive Benchmark Survey is available at randybeandata.com, along with nearly 300 articles published in Forbes, Harvard Business Review, MIT Sloan Management Review, and the Wall Street Journal.

Your organization probably has the right technology. The more complicated question is whether your culture and leadership structure will let you use it.

Listen to the full conversation with Randy Bean on the Data Faces Podcast.

Based on insights from Randy Bean, Senior Advisor and Board Member, featured on the Data Faces Podcast.

Podcast Highlights - Key Takeaways from the Conversation

Podcast highlights

0:53 — Randy shares his unexpected path into data, starting with degrees in English, History, and Art History before training as a COBOL programmer.

1:23 — The “furnace” story: Randy asks a bank manager what they do with customer data and learns they hold it for six years then destroy it.

2:42 — Randy describes flying to Microsoft the same afternoon he was called, walking into a room with Steve Ballmer.

3:55 — Origin of the AI and Data Leadership Survey: a JP Morgan CIO in 2012 asked Randy to survey Fortune 1000 executives on big data.

5:19 — CDO role evolution: from 12% adoption in 2012 to 90% in 2025, and from 28% saying the role was successful in 2020 to 70% today.

8:04 — Diana Schulthaus from Colgate-Palmolive: “I spend 100% of my time on offense.”

10:51 — The 94/6 split: 94% of executives cite culture and people as the principal barrier to AI adoption, only 6% cite technology.

10:51 — Legacy company math: 90% of Fortune 1000 are legacy firms who only need to compete against each other, not the “move fast and break things” 10%.

12:15 — Pandemic transformation: “We’ve done more in six months than in the previous 20 years.”

14:22 — AI at scale progression: 5% in 2023, 24% in 2024, 40% in 2025.

14:22 — “Forget about the FOMO... step back and think about where are we going as a business.”

21:43 — Randy’s HBR article advocating for a unified Chief Data, Analytics, and AI Officer sparked fierce debate with 70% agreement and 30% thoughtful pushback.

21:43 — JP Morgan Chase’s CDO sits on the 14-person operating committee, reports to Jamie Dimon, and was previously the Global Chief Investment Officer.

25:30 — The COBOL programmer lesson: “They’re the people that employ us... maybe we should give them some credit and figure out how to speak their language.”

27:59 — Randy biases “hugely towards an understanding of the business” when selecting data leaders, but notes technical grounding helps you “separate the nonsensical from the realities.”

29:02 — “You gotta know enough to call BS when someone’s giving you a pitch.”

30:27 — The 5-question, 5% framework: “Not all data is created equal, and sometimes 5% of the data is all that it takes to answer 95% of the questions.”

31:44 — The notebook test: Randy describes folding up his notebook ten minutes into meetings when organizations say “we’ve got everything all figured out.”

34:08 — Where to find Randy’s work: randybeandata.com with nearly 300 articles from Forbes, HBR, MIT Sloan Review, and Wall Street Journal.

About David Sweenor

David Sweenor is an expert in AI, generative AI, and product marketing. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter@DavidSweenor and connect with him on LinkedIn.

After years of pilots, proofs of concept, and bold promises about transformation, 2026 is shaping up to be the year when AI investments face real scrutiny. Valuations are wobbling, executives want to see returns, and most organizations still can’t measure whether their AI initiatives are actually working.

Tom Davenport has watched this pattern before. He’s seen technologies overpromise, watched companies chase hype rather than value, and studied what separates the organizations that deliver real results from those that don’t. His message for the year ahead is direct: the quest for value is the biggest thing, and most companies aren’t yet set up to capture it.

About the speaker

Thomas Davenport is a Distinguished Professor at Babson College and one of the most respected voices in analytics and AI. He has spent decades helping leaders turn data and technology into real business value, from coining the term “business process reengineering” to authoring the influential Competing on Analytics. His research and teaching have shaped how companies think about AI, automation, and the evolving impact of generative AI.

In this conversation, we discuss:

• Why we’re in an AI bubble and what that means for enterprise leaders

• The shift from “broad and shallow” pilots to “deep and narrow” implementations

• Why boring transactional use cases will deliver value before transformational ones

• What the P&G experiment with 776 people reveals about disciplined AI experimentation

• The “work slop” problem and why 80% of generative AI output never gets reviewed

• How to build the organizational discipline that separates AI success from theater

The value gap AI needs to close in 2026

“I do think we’re in a bubble,” Tom says. “And I think there’s a lot of real value to AI, of course, but I think the valuations of some of the companies are somewhat crazy.”

Generative AI has attracted a disproportionate share of attention, in part because it’s so accessible. Anyone can use ChatGPT, and anyone can form an opinion about it. That accessibility has inflated expectations beyond what the technology currently delivers. “I think because it’s so accessible to the public and to the chattering classes, if you will, people who write stuff and do podcasts, it’s gotten way more attention than it deserves in the overall kind of pantheon of AI,” he explains.

The market is starting to recalibrate. “We’re starting to see some cracks in the valuations of some of these companies, and suggestions that maybe we don’t need a data center on every corner by a nuclear power plant next door.”

At the organizational level, a different version of the same problem is playing out. Most companies have taken a “broad and shallow” approach to AI adoption, encouraging employees to experiment with tools like ChatGPT for individual productivity. Tom recently co-authored a Harvard Business Review article with Stanford researchers on the need to shift from individual-level use to enterprise-level implementation. The diagnosis resonated; shortly after, another HBR piece appeared with a blunt title: “Stop Running So Many AI Pilots.”

Broad and shallow adoption makes measurement nearly impossible. One company Tom spoke with has essentially given up on proving productivity gains from individual AI tools and now positions them as an employee satisfaction and retention benefit. It’s a candid admission that the value case remains unproven.

“It’s too hard to measure either the productivity or the quality of output, and hardly anybody does measure it.” — Thomas Davenport, Distinguished Professor, Babson College

Start with transactional, not transformational

When asked about the “killer use case” for agentic AI, Tom’s answer catches most people off guard. “Hold on to your hat,” he says with a hint of sarcasm. “Accounts Payable work.”

Keynote speakers don’t get standing ovations for invoice processing, but that’s precisely why it works. The companies making real progress with AI aren’t chasing moonshots. They’re targeting processes that are repetitive, measurable, and low risk. “Generative AI is quite good at sucking the important data out of invoices and sending a message to some other agent,” he explains.

“That’s not exciting, but it is transactional. And I think a lot of people have not terribly exciting jobs, looking at invoices coming in and extracting the key components.” — Thomas Davenport, Distinguished Professor, Babson College

Trust remains a sticking point even in these straightforward applications. Companies aren’t yet comfortable letting AI agents handle complete workflows autonomously. “In a lot of cases, people don’t trust typical agents, so they send it to Stripe or something that they do trust, actually, to pay it when the time comes.” A pattern is emerging: AI handles extraction and routing; established systems handle execution.

The timeline for agentic AI to mature is longer than many expect. Tom estimates “closer to five years before we have real transactional applications,” more optimistic than Andrej Karpathy’s recent prediction of a decade but still far from imminent. For now, companies are “trying to do things that are not terribly risky or important to see how it goes.”

There’s a human dimension here, too. The workers processing invoices don’t have glamorous jobs, but they represent fundamental roles that AI can meaningfully improve. The opportunity isn’t to eliminate these workers overnight but to free them from the most tedious parts of their jobs.

Customer intent is the emerging opportunity

Accounts payable won’t transform a company’s growth trajectory. But customer-facing functions might.

Tom is working with researchers from Cambridge University on a project examining call center interactions. Their early data points to an underappreciated reality: many service calls are actually sales opportunities in disguise.

Service representatives weren’t hired to sell. They were hired to solve problems and answer questions politely. “The people who do that work are generally there because they can answer nicely customer questions about service, and they’re not very good at selling,” Tom notes. The opportunity outpaces the skills of the people handling the calls.

Generative AI could bridge that gap by detecting customer intent in real time. Rather than expecting service reps to suddenly become salespeople, organizations could use AI to identify when a conversation represents a sales opportunity and route it to the right resource.

“In many cases, calls that come into a call center or contact center are not just about service. In many cases, they are often opportunities to sell those customers more.” — Thomas Davenport, Distinguished Professor, Babson College

The implementation Tom envisions involves a channel that “makes sense of what the customers want, and either sends it to another bot that can do that thing, or sends it to a human that can do that thing.” Traditional boundaries between marketing, sales, customer service, and customer success would blur, unified by a shared focus on understanding and responding to what customers actually need. This research is still early, but it suggests where enterprise AI might deliver its next wave of measurable value.

Experimentation is the discipline that separates success from theater

If one thing distinguishes companies capturing real value from AI, it’s their willingness to run rigorous experiments. Tom identifies “disciplined experimentation” as one of the essential capabilities for generative AI success, and he’s blunt about how rare it is.

The example he returns to is Procter & Gamble. He recently spoke with the head of data science and AI at P&G about an experiment they conducted in new product development, designed to answer a straightforward question: Does generative AI actually help people come up with better ideas?

They tested 776 people and carefully measured the results. “Individuals with generative AI were more productive than teams without generative AI,” Tom reports, “and they came up with a better balance of sort of commercial and innovation-oriented ideas.”

What makes this example stand out isn’t just the findings. P&G tested their hypothesis rigorously and published the results with academic collaborators. “Some vendors do it. Anthropic has tested a few things. But in general, companies don’t do that disciplined experimentation,” he says.

“The key thing is that they tested it and ended up writing a paper about it with a bunch of academics. That just doesn’t happen very often.” — Thomas Davenport, Distinguished Professor, Babson College

Most organizations skip this step entirely. They deploy AI tools, encourage adoption, and assume value is being created without any structured way to verify it. AI initiatives are everywhere, with no reliable way to tell which ones are working.

AI work slop is flooding the enterprise

Even when AI delivers useful output, organizations face a stubbornly difficult challenge. A new term has emerged to describe it: “work slop.”

Work slop is the low-quality content that floods organizations when people accept AI outputs without review or refinement. Tom cites a McKinsey study suggesting that 80% of generative AI output is never reviewed. People create the content but never evaluate whether it’s accurate, relevant, or good.

Fixing this requires substantial behavior change. “People have to learn how to get rid of that work slop and edit the output and add some value to the output of generative AI,” Tom says.

He sees the challenge firsthand in his teaching. “I find it really challenging to do with my students, because people seek the easiest way to do something, and we’re not necessarily all trained to be effective editors of content, rather than producers of good first drafts.” Evaluating and improving AI-generated content requires skills different from those needed to create content from scratch, and most people haven’t developed them.

Some of his students have been candid about the tradeoff. They told him, “It was easier to just paraphrase a Wikipedia article” than to iterate on AI outputs, check sources, and add original thinking. That shortcut mentality is understandable, but it undermines the potential value AI could deliver.

“We’re not necessarily all trained to be effective editors of content, rather than producers of good first drafts.” — Thomas Davenport, Distinguished Professor, Babson College

His approach in the classroom offers a model for organizations. “Show me all the prompts you tried,” he tells students. “Show me the edits that you made to the output. Check the sources.” Without that discipline, AI becomes a machine for producing mediocre content at scale.

The path forward

Tom’s perspective on AI in 2026 isn’t pessimistic, but it is grounded in decades of watching technology hype cycles come and go.

“AI can really transform the process of thinking about how you do your work,” he says. That transformation won’t come from deploying more tools or running more pilots. It requires examining workflows, designing experiments, measuring outcomes, and building new habits around how people interact with AI-generated content.

The iterative mindset matters too. “Never take the first output,” he advises. “Iterate on it, ask for alternative interpretations.” Most people accept the first response as definitive when it should be treated as a starting point.

“Doing things the right way is never easy. I guess that’s the lesson here.” — Thomas Davenport, Distinguished Professor, Babson College

The organizations that capture value from AI in 2026 won’t be the ones with the most advanced models or the largest infrastructure investments. They’ll be the ones willing to go deep rather than broad, to measure what matters, and to build the organizational muscle to turn raw AI output into something worth using.

The companies that keep running shallow pilots while waiting for AI to magically deliver value will find themselves in the same place a year from now, still waiting. The ones that do the work won’t have to.

Listen to the full conversation with Tom Davenport on the Data Faces Podcast.

Based on insights from Thomas Davenport, Distinguished Professor, Babson College, featured on the Data Faces Podcast.

Share B2B Marketing Prompts by TinyTechGuides

Leave a comment

Podcast Highlights - Key Takeaways from the Conversation

[2:09] What’s real and what’s hype in AI heading into 2026

[4:24] Why Tom believes we’re in an AI bubble and what needs to come down to earth

[5:51] The unsexy “killer use case” for agentic AI: Accounts Payable

[7:26] Why Tom calls generative AI “predictive analytics on LSD” and his case for the term “analytical AI”

[9:15] The shift from individual-level AI usage to enterprise-level implementations

[11:21] How customer-facing functions represent the next frontier for AI value

[13:26] The disciplines that separate generative AI success from failure

[14:44] Inside the P&G experiment: how 776 people tested whether AI improves ideation

[15:45] The “work slop” problem and why 80% of AI output never gets reviewed

[17:33] Why every prediction about AI and jobs has been wrong

[22:20] How Tom teaches students to use AI the right way: “Show me your prompts, show me your edits”

[24:53] The de-skilling risk: when Tom’s doctor got caught using AI

[30:58] Why business process re-engineering is making a comeback, enabled by AI

About David Sweenor

David Sweenor is an expert in AI, generative AI, and product marketing. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter@DavidSweenor and connect with him onLinkedIn.

2025 was supposed to be the year AI finally worked on its own. We were promised autonomous agents and decision-making without the inconvenience of pesky humans. But, after 27 conversations on the Data Faces podcast, the core message wasn’t about technology at all. It was about us.

In this day and age, it’s hard to get anyone to agree on anything. But after 27 conversations with thought leaders and practitioners, we did agree on one thing. It’s not the technology that is causing AI failures, well, mostly. What was clear was how broken our teams, alignment, and leadership already were. Having been through may of these cycles before, this isn’t new or novel.

“90% of Gen AI projects will fail to deliver transformative value. It’s not that the technology isn’t ready—most organizations aren’t.” — Kjell Carlsson, Domino Data Lab

This year, we talked to analysts, practitioners, and founders from Domino Data Lab, Dataiku, Ernst & Young, Monte Carlo, Relevance AI, Posit, BARC, and more. After 27 conversations, three patterns kept repeating.

Strategy before technology

The leaders who succeed don’t start with tools. They start with business outcomes. Not “let’s implement AI,” but “let’s solve this specific problem.” As Eric Kavanagh put it, the “ready-fire-aim” approach explains most of those failure rates.

“Leaders align AI outputs to corporate KPIs. That tethering of what the corporation wants to succeed with and what AI can help them do—that separates winners from laggards.” — Shawn Rogers, BARC

Agents need guardrails

Agentic AI is already here. It’s handling lead management, maintenance scheduling, and sales enablement. But governance is still a bit nascent. The companies that were doing it previously, such as financial services, healthcare, etc., are further ahead than most. It’s anticipated that some organizations may soon manage 10,000 or more agents. The companies getting this right aren’t moving faster. They’re treating agents like employees, with policies, oversight, and accountability.

“Wrap agents with employee-level policies. Ironically, agents may end up being more compliant than humans ever were.” — Sanjeev Mohan, SanjMo

The human element is the foundation

AI amplifies what’s already there. If your team isn’t aligned, your data isn’t trusted, or your messaging is generic, AI won’t fix it. It’ll expose it.

“If the team is not aligned beforehand, there’s no way whatever model you choose will be successful. Alignment is absolutely vital.” — Danny Stout, Ernst & Young

The same message came through in conversations about ethics, data quality, and messaging. Monica Cisneros pointed out that there are 21 mathematical definitions of fairness. Someone has to choose. Kevin Petrie was blunt. “If the quality ain’t good, the AI ain’t good.” And Emma Stratton warned of the curse of knowledge. Experts assume everyone understands what they do, then wonder why their messaging falls flat.

What 2025 actually taught us

The more powerful AI becomes, the more human skills matter. AI didn’t replace leadership, judgment, and clarity. It made the gaps impossible to ignore. That’s not a technology problem. That’s a mirror.

Hear the full conversations

This is the highlight reel. The full episode goes deeper. 27 guests on strategy, agents, ethics, and what’s coming in 2026.

27 leaders. 3 themes. 1 episode.

Listen to the Data Faces 2025 Year in Review.

🎧 YouTube | Spotify | Apple Podcasts

Listen to the full conversations on YouTube | Spotify | Apple Podcasts

Share B2B Marketing Prompts by TinyTechGuides

Leave a comment

Based on insights from 27 leaders featured on the Data Faces Podcast.

1. The Faces Behind Data: AI, Ethics, and Leadership with Monica Cisneros Data Faces Podcast with Monica Cisneros Nov 14, 2024 | Read more

2. Past as Prologue with Kevin Petrie: What History Tells Us About the Future of AI Dec 11, 2024 | Read more

3. AI in 2025: Why 90% of Gen AI Projects Will Fail | Kjell Carlsson Most AI failures aren’t about the technology—they’re about strategy, governance, and execution. Jan 2, 2025 | Read more

4. Solving the Data Trust Crisis with Kamal Maheshwari Kamal Maheshwari on solving the data trust crisis, aligning teams, and building AI-ready data ecosystems. Jan 21, 2025 | Read more

5. Beyond the AI Hype: What 20% of Companies Get Right Shawn Rogers breaks down BARC’s latest research on AI maturity, the importance of data quality, and what truly separates leaders from laggards. Feb 11, 2025 | Read more

6. AI Agents: State of the Union with Sanjeev Mohan Exploring the current state of AI agents, their challenges, and how businesses can prepare for widespread adoption. Feb 25, 2025 | Read more

7. The AI-Powered CFO: Why Finance Must Shift from Control to Cognition Jawwad Rasheed explains why AI, automation, and self-service analytics are transforming the role of finance leaders. Mar 11, 2025 | Read more

8. There Is No Post-AI World: Preparing Your Organization for the Agent Revolution John Thompson on the strategic value of AI agents, the unexpected risks of autonomous systems, and why intelligent governance matters more than ever. Mar 25, 2025 | Read more

9. The Gen AI Shift: How Product Marketing Managers Are Adapting Melissa Burroughs on scaling PMM productivity, balancing AI efficiency with messaging alignment, and preserving the human element in marketing. Apr 8, 2025 | Read more

10. The Grandmother Test: Building AI Trust Beyond Technology Robert Lake on asking the three essential business questions, managing the human tendency to anthropomorphize, and leading effective AI change management. Apr 22, 2025 | Read more

11. The Customer Hero Principle: Why Your B2B Messaging Falls Flat Gabriela Contreras on ruthless audience prioritization, escaping jargon land, and making customers the center of your product story. May 6, 2025 | Read more

12. Team Dynamics Over Technology: The Human Elements that Drive AI Success Danny Stout on human-centered AI teams, the myth of bigger models, and why communication skills trump technical prowess. May 20, 2025 | Read more

13. From “AI-Ready” to AI Reality: Why Actionable Data Strategies Beat Endless Planning Shane Murray on AI-ready data, the truth about RAG, and why building beats planning for trustworthy AI. Jun 3, 2025 | Read more

14. Stop Chasing Hallucinations—Focus on Agentic Quality Insights from Hyoun Park, CEO & Principal Analyst at Amalgam Insights, on fixing context gaps and building dependable AI agents. Jun 17, 2025 | Read more

15. The “Survival of the Nimblest” Strategy for AI Marketing Success Insights from Judit Szabo, Global Head of Demand Generation at Endava, on balancing automation with human connections in B2B marketing. Jul 1, 2025 | Read more

16. How AI Killed Traditional Competitive Analysis Insights from David Bryson, Principal Competitive Intelligence Manager at Splunk, on turning AI information overload into a strategic advantage. Jul 15, 2025 | Read more

17. How 3% of Companies Win with AI While 97% Fail Rich Mendis reveals the two misconceptions killing most enterprise AI projects and the proven framework that delivers ROI. Jul 29, 2025 | Read more

18. How to Write Punchy B2B Messaging That Actually Converts Emma Stratton from Punchy reveals the curse of knowledge killing most B2B conversions and the proven messaging framework that makes prospects say “yes.” Aug 12, 2025 | Read more

19. The AI Agent Mistake 90% of Marketing Leaders Are Making Chelsea Wise from Relevance AI reveals why learning together beats rushing to implement AI agents and the unsexy use cases that deliver real results. Aug 26, 2025 | Read more

20. Why Bad AI Governance Kills 95% of Enterprise Projects Before Production Thomas Been from Domino Data Lab explains why governance accelerates AI deployment by 70% and the validation trap that kills most enterprise projects. Sep 9, 2025 | Read more

21. Escape the Marketing Twilight Zone: The Agentic AI Playbook for B2B Marketers Rajeev Kozhikkattuthodi from Poexis reveals the three failure modes that prevent marketing teams from moving beyond analysis paralysis to measurable pipeline. Sep 23, 2025 | Read more

22. Your Netflix Moment: Why CIOs Must Act Now on AI Agents Catalina Herrera from Dataiku reveals why most AI agent pilots fail and the four-pillar framework that turns experimental projects into production systems. Oct 7, 2025 | Read more

23. Your AI Project Will Fail. Here Are the Only Three Decisions That Matter AI analyst and DMRadio host Eric Kavanagh on the three unglamorous decisions that separate AI success from expensive failure. Oct 21, 2025 | Read more

24. Augmented Intelligence: The Future of Sales Enablement LaunchDarkly’s Matt Magne shares why augmented intelligence beats automation in sales enablement. Nov 4, 2025 | Read more

25. The Barcode on the Bronze: Why Your AI Needs to Know What Makes You Different Adesso Associates’ Gina von Esmarch reveals how teaching AI your context beats generic automation. Nov 18, 2025 | Read more

26. Data Lineage for AI: Why Truth Beats Hope in Banking Insights from Tina Chace on ensuring data quality in AI deployments. Dec 2, 2025 | Read more

27. Why Code-First Data Science Still Wins in the Age of AI Posit’s Bruno Trimouille explains why governance and innovation aren’t a zero-sum game for data science teams. Dec 16, 2025 | Read more

About David Sweenor

David Sweenor is an expert in AI, generative AI, and product marketing. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter@DavidSweenor and connect with him on LinkedIn.

What happens when your organization makes a major decision based on analysis that no one can verify? The promise of AI-generated code and low-code tools is speed, but speed without trust is just faster failure. Bruno Trimouille, CMO at Posit, makes the case that trustworthy data science still requires code you can inspect and reproduce, and that AI, when used with the proper guardrails, can actually accelerate this approach rather than undermine it. In a recent conversation on the Data Faces Podcast, Bruno shared how Posit thinks about the tension between moving fast and maintaining trust, and how his own team applies these principles in practice.

About the speaker

Bruno Trimouille is the Chief Marketing Officer at Posit, the company formerly known as RStudio. Posit’s mission is to create open source software for data science, scientific research, and technical communication. Today, the company serves 10,000 customers, including 1,800 of the largest firms in regulated industries, and supports millions of users worldwide each week.

Why code-first still matters

Bruno Trimouille didn’t take the typical path to CMO. He started as a software engineer in France, moved through presales and consulting roles serving demanding customers across industries and geographies, and eventually landed in marketing because he saw it as the place where he could combine his technology background with his sales experience to amplify good messaging and storytelling. That engineering foundation shaped how he evaluates opportunities, and it’s what drew him to Posit in the first place.

What attracted Bruno wasn’t the marketing challenge. It was Posit’s foundational belief that trustworthy data science requires code you can question, validate, and repeat. He points to the work of John Chambers, creator of the S language that preceded R, who argued that reliable software must be verifiable and trustworthy. For Bruno, this principle translates directly into how data science should work.

“For software to be reliable, things need to be verifiable, and things need to be trustworthy. So for us, it means that the concept of data science being inspectable and reproducible, thus trustworthy, is paramount.”

— Bruno Trimouille, CMO at Posit

Low-code tools offer convenience, and they’ve helped more people access data science capabilities. But when those tools don’t do what you need, you’re stuck. Code-first isn’t about gatekeeping or making things more complicated for business users. It’s about creating outputs that someone else can examine, challenge, and rerun to confirm the results hold up.

AI as the bridge, not the replacement

Bruno sees the same tension play out in nearly every customer conversation. Data scientists want speed. They want to access data quickly, model it, and share outcomes with business stakeholders before the moment passes. Business stakeholders want speed, too, because the whole point of data science is to make faster, better decisions. And somewhere in the middle sits IT, charged with imposing governance on a process everyone else wants to accelerate.

For years, this tension created a divide. Code-first tools served developers well but felt inaccessible to business users. Low-code platforms promised analytics democratization, but that promise has largely failed to launch. Neither approach served both groups particularly well.

Bruno believes AI is starting to dissolve this standoff, not by replacing code-first approaches, but by making them faster and more accessible.

“AI sort of creates a middle ground approach, which is still code-first based. It bridges the gap, especially when it comes to speed and quick turnaround time, and brings this code-first approach much closer to business stakeholders.”

— Bruno Trimouille, CMO at Posit

The key distinction Bruno draws is between AI that generates black-box outputs and AI that generates artifacts you can actually examine. Posit recently unveiled a set of AI-driven capabilities, and the team is deliberate about calling their approach “responsible AI.” For Bruno, that phrase isn’t marketing language. It means ensuring that everything AI generates is a piece of code, a SQL query, or another technical artifact that can be inspected and rerun for reproducibility.

AI governance isn’t about slowing AI adoption or treating it with suspicion. It’s about recognizing that speed without governance creates liability, while governance designed correctly doesn’t have to slow anyone down.

The three-layer model for scalable, responsible data science

When asked about the common complaint that governance slows things down, Bruno offers a direct rebuttal. He doesn’t see innovation and governance as competing forces. He sees them as two sides of the same coin.

“This is not a zero-sum game between innovation and governance, but rather a framework where governance enables innovation.”

— Bruno Trimouille, CMO at Posit

Bruno describes Posit’s approach as three layers that build on each other, moving from the inside out.

Layer one: Centralized, secure data foundation

Data science starts with data, and security, lineage, and access control at the platform level are non-negotiable. This is why Posit partners with companies like Amazon, Databricks, and Snowflake, because governance at the data layer has to be airtight before anything else can work.

Layer two: Code-first workflow

This layer goes beyond just the model to encompass the entire process of building it. The workflow needs to be repeatable, transparent, and well-documented. Bruno is emphatic on this point, noting that transparency is “really non-negotiable” for many use cases and customers, particularly those in regulated industries.

Layer three: Agile deployment and real-time feedback

When data scientists can instantly deploy models and expose them to business stakeholders, feedback happens in real time rather than weeks later. Stakeholders can interact with the output, run scenarios, and respond while the context is still fresh. As Bruno puts it, “Things can happen in real time, and that pretty much unlocks things.”

The FDA as a proof point

The FDA now accepts clinical trial submissions in open source formats, a significant shift for such a heavily regulated agency. Bruno calls this “a really quantum step by a very established and very governed agency to look at that as new pathways.” He notes that the shift also addresses a practical talent challenge, as skills in legacy, proprietary tools are becoming harder to find. Open source lets these organizations tap into a broader talent pool while maintaining the transparency and reproducibility that regulators require.

From models to mission-critical applications

Data science outputs at Posit’s customers aren’t just models feeding dashboards or reports that sit in someone’s inbox. They’re interactive applications where business stakeholders engage directly with the analysis to make consequential decisions. Bruno describes a spectrum of “data-driven assets” that organizations are building, ranging from APIs that let other systems call models programmatically to scheduled reports with embedded insights to fully interactive applications.

The NASA example stands out.

“If you look at an institution like NASA, they have interactive applications where they look at their staffing needs, staffing prediction for really complex missions like going back to Mars—and interactively play different scenarios, drill into the data, do what-if analysis, and really interact with the data to drive better decisions.”

— Bruno Trimouille, CMO at Posit

What connects these examples is that business stakeholders aren’t waiting for a PDF summary or a static slide deck. They’re working directly with data science outputs, exploring scenarios and stress-testing assumptions to inform real commitments of resources and time. That direct interaction depends on trust, which is precisely why Bruno keeps returning to the code-first foundation.

Practicing what we preach: AI in Posit’s own marketing

Posit sells the philosophy of responsible, inspectable, code-first data science to its customers. But does the company practice what it preaches internally? Bruno’s marketing team offers a valuable test case.

Bruno describes his own evolution with AI as a series of distinct phases. Early on, he saw it as a productivity tool to eliminate “blank page syndrome.” That was valuable, but it was just the starting point.

“I really saw AI as a fantastic tool at the beginning to remove the blank page syndrome... But then I really saw the power that this could deliver in not just being a bot sitting next to you, but literally a thinking partner.”

— Bruno Trimouille, CMO at Posit

As a thinking partner, AI has transformed how Bruno approaches market research. He describes having “market studies on tap,” with the ability to customize research to data science within a specific industry, examining trends and opportunities in ways that would have taken weeks with traditional methods.

On the campaign side, Posit has experimented with mass personalization that maintains clear guardrails. Rather than letting AI write emails from scratch, the team uses templates with defined placeholders that AI can customize based on persona, use case, and industry. Bruno has seen conversion rates of 10 to 12 percent on targeted segments of previously dormant leads. The AI reawakened them through personalization, but within a governed structure.

Throughout these applications, Bruno emphasizes that preserving authentic voice remains essential. As he puts it, “I wouldn’t advocate to have AI generate something, and you press the publish button, and it goes.” The pattern mirrors what Posit advocates for data science more broadly. AI generates, humans inspect.

The hybrid skill set for what comes next

Bruno’s career path from software engineering through presales and consulting to CMO illustrates the hybrid profile he believes will define the next generation of data-driven leaders. He isn’t theorizing about what works. He’s living it.

When asked what skills will matter most going forward, Bruno starts with curiosity. Not curiosity as an abstract value, but curiosity as a practice. He experiments with tools like NotebookLM to turn written documents into audio learning content, recognizing that different people absorb information in various ways.

“Marketing has become a greater mix between the art and the creative stuff, but science and data. You need to have a more hybrid skill set. Think about AI as a thinking partner.”

— Bruno Trimouille, CMO at Posit

Bruno also points to cross-functional collaboration as essential rather than a nice-to-have. He now partners regularly with Posit’s data science team to develop market and data insights, a partnership that simply didn’t exist five years ago. Marketing as a discipline has become what he calls “a team sport,” and the teams you need to play with keep expanding.

There’s also a systems dimension to leadership that Bruno emphasizes. He thinks of his marketing stack as an architecture to ensure tools are connected and data flows smoothly. Leaders who treat their technology as a collection of disconnected point solutions will struggle compared to those who design for integration.

Bruno closes with an observation that applies well beyond marketing.

“Having raving fans beats all kinds of marketing campaigns you can put together.”

— Bruno Trimouille, CMO at Posit

The lesson for data science leaders is the same. Build trust with your stakeholders through transparency and reliability, and you earn credibility that no dashboard or presentation alone can deliver. The code-first philosophy Bruno advocates isn’t just about technical rigor. It’s about building the kind of credibility where people believe what you show them because you’ve never given them a reason not to.

Listen to the full conversation with Bruno Trimouille on the Data Faces Podcast.

Based on insights from Bruno Trimouille, Chief Marketing Officer at Posit, featured on the Data Faces Podcast.

Leave a comment

Share B2B Marketing Prompts by TinyTechGuides

Podcast Highlights - Key Takeaways from the Conversation

0:52] Bruno introduces Posit, its rebrand from RStudio, and the company’s mission to create open source software for data science, scientific research, and technical communication.

[2:06] Bruno shares his unconventional path to CMO, starting as a software engineer in France and moving through presales and consulting before landing in marketing.

[4:53] The case for code-first data science: Bruno explains how John Chambers’ principle that reliable software must be verifiable shaped Posit’s foundational philosophy.

[7:32] Speed meets governance: Bruno describes the tension between data scientists wanting velocity and IT demanding control, and how responsible AI can bridge that gap.

[9:13] The three-layer model: Bruno outlines Posit’s framework for scalable data science, from secure data foundations to code-first workflows to real-time deployment and feedback.

[11:29] From models to mission-critical apps: Bruno shares how NASA uses interactive applications for staffing predictions on Mars missions, running what-if scenarios in real time.

[15:53] The FDA’s “quantum step”: Bruno discusses how the FDA now accepts clinical trial submissions in open source formats, and why this matters for talent and transparency.

[18:09] Community as strategic moat: Bruno explains why he had to shift his mindset from seeing community as a marketing channel to treating it as an extension of the team.

[21:31] AI as thinking partner: Bruno describes his evolution from using AI to fix “blank page syndrome” to having “market studies on tap” for strategic research.

[24:04] Mass personalization with guardrails: Bruno shares how AI-driven email customization achieved 10-12% conversion rates on dormant leads while maintaining governed templates.

[34:04] The hybrid skill set: Bruno on why curiosity, cross-functional collaboration, and systems thinking define the next generation of data-driven leaders.

About David Sweenor

David Sweenor is an expert in AI, generative AI, and product marketing. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter@DavidSweenor and connect with him on LinkedIn.

When you see a dashboard or report in your BI system, do you question whether it’s actually correct? Most people don’t, and assume the data flowing through pipelines must be accurate. That assumption holds until the numbers fail to reconcile.

Tina Chace made that same assumption early in her career. Starting in middle office operations, booking trades, she learned that “the importance of making sure that your data is absolutely correct, especially when it comes to transacting and dealing with money, was drilled into me from the start.” But she didn’t fully appreciate what that meant until she spent six years deploying AI and machine learning models in highly regulated environments, specifically transaction monitoring and Know Your Customer systems for major banks.

The pattern she discovered changed everything.

“I didn’t really start to appreciate the value of data and accuracy and trust in your data until I spent six years working at an AI and machine learning company, specifically rolling out machine learning models in highly regulated spaces...I was continuously running into problems, and 90% of the time it ended up being a data issue.”

— Tina Chace, VP Product Management, Solidatus

About Tina Chace

Tina is Vice President of Product Management at Solidatus. She started in middle-office operations, moved to deploying AI models for major banks, and discovered the hard way that 90% of production problems can be traced to data quality issues. That journey from assuming data worked to proving it works shapes her approach to lineage. While tracking technical details is essential, organizations must never lose sight of the business consequences.

From Naive Trust to Earned Skepticism

Tina’s journey from naive trust to earned skepticism mirrors what most data leaders experience. Working in the middle office where financial transactions required absolute precision, she assumed someone, somewhere, was ensuring data quality. Information moved through pipelines, and transactions were being processed. Her assumption was simple. “This data is of high quality, and we know what’s happening to it.”

Deploying machine learning models shattered that assumption. Someone would make a change upstream and add a new field with sensitive data. Undetected, this data would flow into the model. Only after days of troubleshooting would she identify the culprit: an unexpected upstream change.

Why Knowing More Makes You Trust Less

Tina describes an oddity. “The more you know, the more you almost distrust, right? Like the more educated you are, the more you question your position.”

Five years ago, teams were largely unaware of how data moved through their organizations. Today, they can see the complexity across dozens of systems and hundreds or thousands of transformations. But visibility without control just amplifies anxiety.

“People were more unaware previously, five years ago, and now there’s more visibility, but also with more visibility, there can be a bit more anxiety, because then, you know, it’s not covered.”

— Tina Chace, VP Product Management, Solidatus

The stakes vary by context. Sending marketing coupons carries a different risk than processing financial transactions. But the same questions surface. Who verifies report accuracy? What happens when a CFO reviews two reports that should show identical revenue figures? One says $47.3M. The other says $47.5M. Finance blames data engineering. Engineering points to the source system. Three days later, nobody knows which number is correct.

Consider rounding errors, similar to those depicted in the movie Office Space. In small transactions, a rounding discrepancy to the wrong decimal place barely registers. “When you’re dealing with large amounts of money, that rounding error becomes a huge financial burden if it is incorrect.” Contracts between applications can specify different rounding rules without anyone documenting the difference. Left unchecked, the discrepancy silently compounds across every transaction.

Regulatory requirements like BCBS 239 now mandate that organizations “report on certain metrics and ensure the data going into that report is accurate and timely and correctly calculated.” Assumptions no longer suffice, and auditors demand proof.

These trust gaps don’t emerge randomly. They cascade from organizational failures that compound with every system the data touches.

Three Failures That Compound

Consider what happens when a data transformation changes upstream. The data engineer implementing the change often lacks visibility into downstream reports and applications, failing to notify stakeholders. The analyst who owns the report doesn’t speak the engineer’s technical language, so they can’t diagnose the problem when discrepancies appear. And because the transformation was never documented, troubleshooting takes days instead of hours. Three organizational failures, each enabling the next.

Here’s how the cascade works.

First Breakdown: The Ownership Vacuum

Ask your organization this question. Who owns the quality of customer data from CRM entry through analytics warehouse to the executive dashboard? You’ll get different answers from every team.

Ownership is distributed across “various points of its life cycle,” which often means no one owns it. Data engineers build pipelines, analysts create reports, and business leaders consume outputs. Each assumes someone else ensures accuracy. Most organizations can’t point to a single person accountable for end-to-end data quality from the source system through the final report. The gaps between ownership zones are where problems hide.

Second Breakdown: The Language Barrier

Without clear ownership, teams can’t communicate effectively. Engineers discuss systems and ETL processes. Analysts talk about reports and metrics. Executives focus on outcomes and risk—three different vocabularies describing the same data, with no shared language to bridge them.

Technical teams understand data flows but struggle to communicate the implications to business stakeholders or reporting analysts. Engineers know WHAT is happening. Business teams need to understand WHY it matters. Without translation, that gap never closes.

“If you, from a technology standpoint, understand your data flows, but you cannot communicate the meaning of that or the implication of that to your business stakeholders or the reporting analysts looking at the end report, then you’ve got a communication gap.”

— Tina Chace, VP Product Management, Solidatus

Third Breakdown: The Documentation Death Spiral

When nobody owns the whole picture, teams can’t communicate effectively. Without ownership, documentation becomes obsolete immediately. What gets documented? By whom? For whose benefit?

Tina’s troubleshooting experience shows the cost. Finding the root cause takes days, but why? Because transformations aren’t tracked and calculations aren’t recorded. Quality checks run in some systems but not others. And upstream changes propagate downstream invisibly, often only being discovered when something breaks.

Point solutions that map individual systems fail to solve organizational breakdowns. That requires a shared foundation where engineers, analysts, and executives work off the same set of blueprints.

Two Dimensions, One Foundation

Solidatus operates on the premise that technical tracking alone is insufficient. Organizations need both technical and business lineage.

• Technical lineage tracks “where a column, such as trade date, is flowing through various applications before it gets booked in a report.” It maps the mechanical flow: specific systems, transformations, and connections.

• Business lineage captures “the quality of your data as it’s flowing through systems, what kind of controls or checks are happening in various systems, and who owns the data at various points of its life cycle.” It answers why this flow matters and who’s accountable at each stage.

Without both, organizations either track data but can’t communicate implications, or discuss business concerns but can’t trace them to technical root causes.

“We at Solidatus think it’s important to have both technical and business lineage...not just tracking where it flows or any kind of calculations, but understanding context, such as the quality of your data as it’s flowing through systems, what kind of controls or checks are happening, and who owns the data at various points of its life cycle.”

— Tina Chace, VP Product Management, Solidatus

Every Transformation, Every System, Every Step

Column-level lineage captures every transformation that data undergoes. At each step and with each application, teams track calculations, rounding rules, and aggregations. By the time data reaches a report, teams are aware of its complete history: every change, every system, every decision point.

This granularity solves the discrepancy problem Tina encountered repeatedly. When two reports should match but don’t, teams can investigate the discrepancy to determine the cause. “We rounded to a specific decimal between applications A and B, but changed the rounding rule between B and C.”

What previously took days of detective work can now be traced in hours.

Proactive vs. Reactive: Shifting Quality Checks Left

Beyond tracking flows, lineage enables quality checks across every single system. Instead of discovering a $200K discrepancy during the board meeting, the quality check flags it at 3 am when upstream rounding logic changes. The data engineer fixes it before anyone downstream notices.

If there’s a failure, teams know immediately which systems are downstream and which specific report fields might be inaccurate. Reactive firefighting becomes proactive prevention.

Governance teams document which checks and policies apply at each stage of the process. Sensitive information either gets obscured or never enters the pipeline. Access controls are enforced at every step, not just endpoints. This is where compliance shifts from aspirational to demonstrable.

Same Blueprint, Different Lenses

Different stakeholders view the data flows through their own lens, all based on the same underlying information. Data engineers see technical flows and transformation logic while analysts understand which quality checks protect their reports. Governance teams track the location of sensitive data and identify the applicable controls. Executives see confidence metrics and risk exposure—all perspectives drawing from the same lineage capture.

This shared foundation solves the language barrier. Engineers and business leaders are no longer translating between different systems. They’re viewing different aspects of the same reality.

This foundation solved problems Tina encountered when deploying models years ago. Now that AI is proliferating across organizations, those problems are multiplying. And the stakes are higher.

Why AI Makes Incomplete Lineage Unacceptable

“With the proliferation and popularity of using AI within companies, I’m even more concerned about understanding the data that flows into it,” Tina explains. Her six years deploying models taught her that incomplete lineage “led to a lot of real-world problems.”

Are You Automating or Just Shifting Where You Monitor?

Demos showcase impressive AI results, but production environments are different. Organizations deploy AI to achieve better metrics, higher productivity, and automated decision-making. But at runtime, teams need confidence in what’s feeding the model.

Without that confidence, the promise of automation loses its appeal. “If I have to monitor it all the time, it didn’t save me any productivity at all. I’m just monitoring instead of doing it manually.”

The decisions AI makes aren’t abstract. They impact mortgage approvals, payment processing accuracy, and the detection of fraudulent transactions. In Tina’s work with Know Your Customer systems at major banks, errors have immediate financial and regulatory consequences.

Privacy Requires Proof

“One of the big concerns about the use of AI is privacy, like, where is my information being used?” Lineage provides attestation. Teams can prove models are “obscuring certain private information that shouldn’t be in it, or it’s not even entering the models at all, unless absolutely necessary.”

Compliance audits shift from defending processes to demonstrating controls. Instead of explaining what should happen, teams show what does happen.

Partial lineage creates blind spots where problems hide. Teams can’t skip systems or steps. Incomplete coverage leaves gaps where transformations remain invisible, precisely where issues emerge during troubleshooting.

Don’t Boil the Ocean

Faced with incomplete or nonexistent lineage, where do organizations begin?

“Be very specific and deliberate on what you’re choosing to address first,” Tina advises. “Then you will have an immediate and tangible win by covering a specific scenario, and you can continue to expand out from there in terms of importance.”

The alternative is “trying to document every single system that you have.” That becomes “this big, nebulous project” where “you won’t have an output until five years from now.” Starting with critical use cases delivers incremental ROI, rather than waiting years for enterprise-wide perfection.

Your First Four Targets

Prioritize data flows that intersect risk and visibility.

1. Regulatory Reporting: Systems that feed BCBS 239 capital reporting or mandated metrics, where auditors require proof of accuracy.

2. AI models are currently in production. Transaction monitoring, KYC, fraud detection, or other automated decision systems where bad inputs can have immediate consequences.

3. Areas with known quality issues. Recent troubleshooting incidents, recurring discrepancies, or pipelines that match that 90% pattern.

4. High-stakes decision systems. Mortgage approvals, payment processing, or other flows where errors have a direct financial or customer impact.

Start where one of these applies. Demonstrate value. Expand based on business priority, not technical completeness.

From Anxiety to Action

The ‘visibility paradox’ Chace describes—where knowing more creates more anxiety—can only be resolved through deliberate action. Knowing more creates anxiety. Seeing problems without the ability to address them compounds stress rather than relieving it.

“Regulations and recommendations are improving the metrics and trust we have in the data we’re using.” Organizations implementing lineage deliberately implement lineage, starting with critical use cases and building incrementally, establish the foundation for trusted AI and confident decision-making.

The alternative is waiting until the next crisis forces action. When a model makes a costly error, when an auditor asks questions nobody can answer, when that $200K discrepancy appears in the board deck.

Next time you see a dashboard, question it. Then ask whether you could trace that number back to its source. How long would it take, hours, days, weeks? Visibility without traceability isn’t insight, it’s anxiety with data attached. The organizations that can answer “yes” aren’t lucky; they’re prepared. They started with one critical pipeline, proved the value, and expanded from there.

The question isn’t whether you need lineage. It’s whether you’ll implement it before or after your next crisis.

Listen to the full conversation with Tiny Chace on the Data Faces Podcast.

Based on insights from Tina Chace, VP Product Management at Solidatus, featured on the Data Faces Podcast.

Share B2B Marketing Prompts by TinyTechGuides

Leave a comment

Podcast Highlights - Key Takeaways from the Conversation

[0:54] Tina’s Career Journey: From Trust to Skepticism

“I actually started my career in the middle office and booking of trades and trade data. So the importance of making sure that your data is absolutely correct, especially when it comes to transacting and dealing with money, was drilled into me from the start.”

Key Takeaway: Early career assumption was that data quality was guaranteed. Reality proved different.

[2:28] The 90% Data Problem Discovery

“I was continuously running into problems where someone would make a change upstream, maybe they added a new field that had sensitive data, and it would flow into the model. We wouldn’t know that, and we’d have to go back and troubleshoot, and 90% of the time it ended up being a data issue.”

Key Takeaway: Six years of deploying AI/ML models in highly regulated spaces (transaction monitoring, KYC) revealed that 90% of production problems trace back to data issues.

[3:24] What is Data Lineage?

“Data lineage allows you to track how your data is flowing through various systems...not just tracking where it flows or any kind of calculations, but understanding context, such as the quality of your data as it’s flowing through systems, what kind of controls or checks are happening in various systems and who owns the data at various points of its life cycle.”

Key Takeaway: Data lineage requires both technical tracking (where data flows) AND business context (quality, controls, ownership).

[6:47] The Rounding Error Problem

“When you’re dealing with large amounts of money, that rounding error becomes a huge financial burden if it is incorrect...between application A and B, we rounded to this decimal between application B and C. We actually changed the rounding, and that actually ended up in our capital reporting.”

Key Takeaway: Small rounding errors compound at scale. Contracts between applications can specify different rounding rules without documentation.

[7:59] Column-Level Granularity

“By documenting data at the most granular level, which we call column level, in the data lineage world, you can actually document for every step and application that that data flows through. Is there a calculation? Is there a rounding? Are you adding things together? So that by the time it ends up in a report, you actually understand exactly what happened for it to be in that report.”

Key Takeaway: Column-level documentation captures every transformation, enabling precise troubleshooting when discrepancies appear.

[8:26] Bridging the Language Gap

“If you, from a technology standpoint, understand your data flows, but you cannot communicate the meaning of that or the implication of that, to say, your business stakeholders, or the reporting analysts who are looking at the end report, then you’ve got a communication gap.”

Key Takeaway: Technical teams know WHAT is happening; business teams need WHY it matters. Lineage bridges this gap.

[9:46] The Shared Blueprints Approach

“It really brings together all the different stakeholders, and they can view their own lens of the data flows, but it’s based on the same underlying information, so you’re working off of like, the same set of blueprints.”

Key Takeaway: Different stakeholders (engineers, analysts, governance, executives) view different aspects of the same lineage data.

[30:57] Why AI Makes Lineage More Critical

“With the proliferation and popularity of using AI companies, using AI within companies, I’m even more concerned about understanding the data that flows into it...I recognize that not having this data lineage and understanding the data flows led to a lot of real world problems.”

Key Takeaway: The deployment of AI amplifies the need for complete data lineage; incomplete visibility creates unacceptable risk.

[32:14] The Productivity Paradox

“At runtime, I want to be sure that everything that’s going into it is helping it make the best decision, because if I have to monitor it all the time, it actually didn’t save me any productivity at all. I’m just monitoring instead of doing it manually.”

Key Takeaway: AI that requires constant monitoring hasn’t actually automated anything—you’ve just shifted where you spend time.

[32:32] Privacy and AI

“One of the big concerns about the use of AI is privacy, like, where is my information being used? With data lineage, you can attest to the fact that your models are obscuring certain private information that shouldn’t be in them, or it’s not even entering the models at all, unless absolutely necessary.”

Key Takeaway: Lineage enables privacy attestation—proving what data enters AI models, not just promising.

[33:26] The Trust Paradox

“The more you know, the more you almost distrust, right? There’s that paradox, like the more educated you are, the more you question your position...People were less aware five years ago, and now there’s more visibility. Still, also with more visibility, there can be a bit more anxiety, because then, you know, it’s not covered.”

Key Takeaway: Increased visibility into data complexity creates anxiety alongside awareness—seeing problems without solutions amplifies stress.

[34:27] Don’t Boil the Ocean

“Be very specific and deliberate on what you’re choosing to access first, because then you will have an immediate and tangible win by covering a specific scenario, and you can continue to expand out from there in terms of importance...You can get ROI and value out of capturing your most critical use cases first...rather than trying to do a boil the ocean exercise, and you won’t have an output until five years from now.”

Key Takeaway: Start with critical use cases, prove value, expand incrementally. Don’t try to document everything at once.

About David Sweenor

David Sweenor is an expert in AI, generative AI, and product marketing. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter@DavidSweenor and connect with him on LinkedIn.

​​Last week, three different B2B companies showed me their AI-generated customer personas. I couldn’t tell which belonged to which company. Neither could they.

This is the hidden cost of AI trained on “the world’s internet”—it…

Read more

Matt Magne has seen this movie before. Twenty years ago, as a sales engineer in the master data management (MDM) space, he watched companies spend millions trying to create a “single view of the customer.” Today, as Senior Enablement Manager at LaunchDarkly, he’s watching the same pattern unfold with sales enablement tools.

“People spent millions of dollars on a solution to do that, and they still haven’t fixed that problem,” Matt reflects about his MDM days. Now he sees sales reps juggling 20 to 30 different applications, each containing a fragment of truth about their performance and capabilities. The irony isn’t lost on him that adding AI role-play tools to this mix, no matter how sophisticated, just creates another silo.

About Matt Magne

Matt describes himself as the “AI-powered Silicon Valley sales enablement guy” and brings an eclectic background to the role. From coder to sales engineer to product marketer, his career spans technical and creative pursuits, including a band featured on MTV’s Road Rules and a guitar-playing TED Talk. Eight months into his role at LaunchDarkly, he’s experimenting with voice-enabled AI for sales practice while maintaining healthy skepticism about technology’s limitations.

Why voice AI works (and why it doesn’t matter without integration)

LaunchDarkly sells feature flags, which Matt explains through his Christmas lights analogy. Install lights once, then flip switches throughout the year to display different colors without climbing back on the roof. Feature flags let developers deploy code once and control features without new releases. This technical complexity creates specific sales challenges when prospects dismiss it as “just another config tool.”

Matt discovered a voice-enabled AI that lets reps practice these difficult conversations. “You can verbally give an objection to it and it will verbally respond,” he explains. The AI simulates skeptical enterprise buyers who challenge pricing and question architecture decisions. Reps practice the same conversation multiple times, experimenting with different approaches without fear of judgment.

“I play guitar. One of my challenges is that I think about playing with the metronome and all that by myself, and I’m terrible at it. The same thing goes for role plays, where it’s like, I’ll prepare with someone I’m comfortable with, but it’s way different than being on stage.”

Matt Magne, Senior Enablement Manager, Revenue Enablement at LaunchDarkly

The technology works remarkably well for creating safe practice spaces. But here’s where Matt’s MDM experience provides a crucial perspective. Just as companies had customer data spread across dozens of systems twenty years ago, today’s sales organizations have performance insights scattered across their tech stack. The AI might capture that a rep struggles with security objections, but if that insight doesn’t reach the right coach at the right time, deals still get lost.

The elephant problem: Everyone’s touching different parts

Matt uses a vivid metaphor to describe the current state of sales enablement. “We’re all touching the elephant at different spots, and one thinks it’s a tree,” he explains. The manager looks at pipeline metrics, the enablement team checks training completion, and the AI tool logs practice patterns. Each stakeholder has valid data but sees an entirely different picture.

“All these SaaS solutions that we have, I mean, some AEs have like 20, 30 solutions they’re using, and there’s still a challenge of data silos, pulling the data together.”

Matt Magne, Senior Enablement Manager, Revenue Enablement at LaunchDarkly

This fragmentation becomes expensive when reps encounter scenarios they haven’t mastered. Matt notes that reps typically care about training “when their tire goes flat”—meaning they scramble for help only after fumbling a critical customer conversation. By then, the opportunity is usually lost.

The path forward isn’t more sophisticated AI but better connections between existing tools. Matt mentions MCP servers as one potential solution, though he remains realistic: “It’s still an integration game.”

Build for augmentation, not automation

Matt’s experiments revealed a counterintuitive truth about AI in sales training. Perfect AI responses actually create worse outcomes than slightly flawed systems. “They’re spooky, right? They’re so smart, but they’re still really dumb in a lot of ways,” he observes. When AI occasionally misunderstands context, reps develop stronger critical thinking skills to handle real customer curveballs.

“I think I read an article about how hallucinations are a mathematical inevitability... You’re still gonna need a human in the loop.”

Matt Magne, Senior Enablement Manager, Revenue Enablement at LaunchDarkly

This philosophy shapes Matt’s vision for sales enablement. Rather than replacing human coaches, AI should surface insights that help managers provide targeted support. Picture AI noticing struggle patterns and automatically alerting the right coach with specific conversation examples and suggested interventions.

Matt recalls discussions from eight years ago about whether AI would replace sales roles. “We were talking about it then, and they still haven’t,” he notes. The future belongs to teams that use AI to make humans more effective, maintaining what Matt calls “someone gluing everything together, and someone kind of making sure it’s not completely off the rails.”

The real playbook for data leaders

Matt’s experience offers three actionable insights for data science leaders evaluating sales enablement AI:

Start with integration architecture, not AI features. Map where performance data currently lives across your organization. Build connections between systems before adding new AI capabilities. The fanciest role-play tool becomes just another silo without proper integration.

Design for human-AI collaboration. The goal isn’t to automate coaching but to surface insights that help human coaches intervene effectively. Slightly imperfect AI that keeps humans engaged produces better results than flawless automation.

Focus on the “flat tire” moments. Reps need help when they’re stuck in real conversations, not during scheduled training. Build systems that deliver relevant practice and coaching exactly when reps will actually use it.

The uncomfortable truth Matt’s experiments reveal is that most sales organizations aren’t ready for AI transformation because they haven’t solved basic data integration. Until you can connect insights from practice sessions to coaching conversations to deal with outcomes, even the most sophisticated AI just adds complexity without solving core problems.

Connect and learn more

Subscribe to the Data Faces Podcast for more conversations with leaders making AI and analytics work in the real world. Available on YouTube | Spotify | Apple Podcasts

Want to discuss implementing voice-enabled AI for sales enablement without falling into the integration trap? Connect with Matt Magne on LinkedIn at linkedin.com/in/exrocker.

Listen to the full conversation with Matt Magne on the Data Faces Podcast.

Based on insights from Matt Magne, Senior Enablement Manager, Revenue Enablement at LaunchDarkly, featured on the Data Faces Podcast.

Podcast Highlights - Key Takeaways from the Conversation

[0:06] The evolution from MDM to AI-powered enablement

Matt shares his journey from solving master data management problems 20 years ago to tackling the same integration challenges in sales enablement today - proving some problems persist across decades and technologies.

[1:53] LaunchDarkly explained: The Christmas lights analogy

Understanding feature flags through a brilliant analogy - install lights once, flip switches all year without climbing back on the roof. This complexity creates unique sales training challenges.

[6:11] Why traditional sales enablement is broken

The “flat tire principle” - reps only care about training when they’re already stuck in a customer conversation, making traditional front-loaded onboarding ineffective.

[10:15] Voice AI breakthrough: Real conversations, not scripts

“You can verbally give an objection to it and it will verbally respond back” - how voice-enabled AI creates safe practice spaces where reps actually want to train.

[21:23] The 30-tool disaster: Sales tech stack fragmentation

AEs juggle 20-30 different solutions daily, creating data silos that prevent coaches from seeing the full performance picture.

[28:30] Why perfect AI is worse than flawed AI

“They’re spooky smart but still really dumb” - Matt’s counterintuitive discovery that hallucinations and imperfections actually improve rep critical thinking.

[33:18] The holy grail: 60-minute sessions condensed to 7 minutes

Matt’s vision for AI that intelligently summarizes long training sessions into digestible content reps will actually consume.

[37:05] Integration before innovation: The MCP server opportunity

Breaking down data silos matters more than adding new AI tools - why the future requires connecting existing systems first.

[40:59] Augmentation, not automation: Keeping humans in the loop

“The future is still augmentation” - why AI won’t replace human coaches but will make them dramatically more effective.

About David Sweenor

David Sweenor is an AI, generative AI, and product marketing expert. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn.

Over the past 20 years, companies spent $2 trillion on Google AdWords—the foundation of how B2B companies get customers to find them online. ChatGPT killed that model in 18 months. The same pattern is emerging with LinkedIn marketing and will likely repeat with AI-based search. Organic inbound traffic, the bedrock of modern B2B marketing, is quickly dying.

“Web traffic for the organic is way down,” says Eric Kavanagh, AI analyst and host of DM Radio. “Companies are freaking out. Google’s empire is teetering.”

About Eric Kavanagh

Eric Kavanagh is an AI analyst and host of DM Radio. Since 2005, he’s conducted over 2,000 podcasts, radio shows, and webinars, including 20+ years with The Data Warehousing Institute (TDWI). He’s watched every major technology wave: the dotcom boom, big data, cloud computing, and now AI. He’s not a futurist making predictions—he’s a pattern-spotter who’s seen this chaos before.

While Google scrambles, most companies aren’t even paying attention to that disruption. They’re too busy making worse mistakes with their own AI initiatives. Eighty to ninety-five percent of AI projects fail because nobody asked basic questions before starting. The technology works fine, but their thinking on how to approach AI doesn’t.

“We’re in this sort of ready, fire, aim mode,” Kavanagh says. “Many companies are pulling the trigger on programs, and that’s why you’re seeing 80% failure rates, 95% failure rates. People didn’t really think through what they were trying to do with this stuff.”

He’s identified three decisions that separate the 20% who succeed from the 80% who burn money and credibility. Get them right and you’ll deliver measurable value while your competitors chase demos. Get them wrong and you’ll join the companies explaining to the board why that $2M AI initiative produced nothing but expensive lessons.

Decision 1: Choose tedious, focused problems over impressive demos

Eric Kavanagh asked the AI agent Manus to book plane tickets for his wife. He watched it work, clicking through sites, checking prices, navigating booking flows. “I was wildly faster doing things the old-fashioned way than Manus was, and Manus came up with a price that was twice as much.” This is the pattern haunting many AI projects; they look impressive in demos, but are useless in practice. Why is it that every agentic demo talks about managing your calendar? Is it really causing you that much grief?

The AI systems that work aren’t very glamorous for demos. They often handle boring tasks like anti-money laundering and regulatory compliance checks. “Doing things in the background or on the side,” Kavanagh says. Success looks like “our system processed 10,000 transactions with 99.5% accuracy,” and not “our business leaders transformed how we think about customer engagement.”

New AI models come out constantly, and new capabilities are announced on a daily basis. AI vendors promise the moon while legacy players AI-wash their capabilities. When everything changes this fast, companies reach for whatever looks impressive instead of figuring out what’s actually useful and strategic. “Take a deep breath,” Kavanagh says. “Figure out, what are we trying to accomplish with these things? What are the cost models?” Most companies skip that step. They want transformation when they need Tuesday’s invoices processed correctly.

The Manus test is simple: Watch your AI work and ask yourself if it’s genuinely better than doing it the old way. If the honest answer is no, you’ve built an expensive demo that costs twice as much and delivers half the value.

Even when you choose focused, boring problems that pass the Manus test, you still face a second decision. How do you actually govern these systems when nobody — including their creators — understands how they work?

Decision 2: Audit logs over guardrails

“Guardrails in general are overrated.”

That’s Eric Kavanagh’s position on AI safety. While Anthropic calls itself “the safety company” and OpenAI reportedly trains models to “only blackmail as a last resort,” Kavanagh thinks most of it is theater. Not because safety doesn’t matter, but because guardrails don’t actually work.

The evidence shows up everywhere. Google’s Gemini refuses to answer anything remotely political which may make it too careful to be useful. Grok gave one user detailed instructions for hacking a smartphone to improve battery life — too loose to provide real protection.

There’s no sweet spot because you can’t build perfect guardrails around systems that nobody fully understands, including the people who created them. “The big models, even the people who designed them don’t know exactly how they work,” Kavanagh points out. Imagine regulators trying to audit OpenAI’s systems. “Think about going into the offices at OpenAI and saying, All right, well, let me see the system you’re using, right? There’s no way, dude, you can’t survey all that.”

Someone tested this opacity recently. They published a webpage with a “don’t index” tag and invented a new word. The next day, ChatGPT found it. Nobody knows how. The systems are doing things nobody can verify or explain. But, that’s par for the course. Why would we expect anything different from these companies with over-inflated valuations?

This is the reality. The systems are opaque and their behavior is unknowable. With billions of parameters and the nature of neural networks, audits are impossible. And despite this, companies are pouring resources into guardrails that either block legitimate work or fail to stop harmful use.

Kavanagh’s alternative strips away the complexity:

“Audit logs. Just audit logs, as long as you’ve got some log file that says what it did where. I mean, that’s what all the AI agent companies are talking about doing. It’ll log what it does, and that way you can go back and watch it. You have to be able to kill the pods basically, that launch whatever structure you have.”

— Eric Kavanagh, AI Analyst and host of DM Radio

The framework comes down to four questions. What did it do? Where did it do it? Can you stop it? If your AI agent starts sending emails to customers or updating financial records, you need to be able to shut it down immediately and not wait for a deployment cycle. Can you fix the damage if something goes wrong?

“That’s basic governance,” Kavanagh says. It’s not sophisticated and it’s not perfect. But, you can actually build it with today’s tools, unlike the fantasy of perfect guardrails.

Guardrails try to prevent disasters that nobody can predict in systems nobody understands. Audit logs accept that reality. They document what happened and give you the ability to respond when things go wrong, albeit, probably too late.

It’s unglamorous governance for systems nobody fully understands. But it’s honest about the trade-offs. And it’s what the 20% who succeed are quietly building while everyone else argues about prompt engineering and fine-tuning.

Audit logs tell you what happened. But there’s a third decision that determines whether you should have let it happen in the first place.

Decision 3: Never mix deterministic with probabilistic systems

Fifty percent error rate! That’s how often AI gets medical recommendations wrong, according to recent studies Kavanagh cites. Chain multiple AI calls together in workflows and the math gets worse. “There’s an error rate at every step of the way, and when you multiply that out, the error winds up being like 40% or something.”

For systems recommending Netflix shows, that’s fine. For systems approving credit, processing transactions, or managing regulatory compliance, it’s lightning in a bottle.

“Databases aren’t going away. Transactional systems aren’t going away. They will be aided and abetted by these other systems, but you have to be careful not to mix those two... For most business decisions, you need to be very sure about what you’re doing.”

— Eric Kavanagh, AI Analyst and host of DM Radio

The danger isn’t AI suddenly replacing your deterministic systems. The danger is subtler. Probabilistic systems slowly become deterministic as humans get tired of reviewing recommendations. The AI didn’t stage a coup and people just got tired of clicking “override.” That’s how you end up with 40% error rates making high-stakes calls.

Here’s how it happens. Your customer support team starts using AI to route tickets. It works well — let’s say at about 90% accuracy. Then someone realizes the AI is pretty good at suggesting refund amounts too. Within six months, support reps stop reviewing the suggestions. Nobody decided to let AI make refund decisions. It just happened. The next thing you know is when you discover the AI has been approving fraudulent refund requests at a 35% error rate for three months. That’s the risk companies face.

Deterministic systems like databases and transaction processors must be certain. They make final decisions with legal, financial, or regulatory consequences. Probabilistic systems like AI are helpful but unreliable. They can be wrong, as long as they’re not making decisions that require certainty.

Here’s where to draw the line. Financial transactions, credit approvals, compliance reporting, and access control need to be deterministic. Support ticket routing, content recommendations, and marketing personalization can be probabilistic.

Kavanagh believes “small language models, or just old-fashioned, deterministic AI models, are going to be ruling the day, at least I hope so, because the big stuff is too big, and it’s unwieldy and you can’t trust it.” The complexity of large models creates unpredictability. Size doesn’t equal performance. It equals loss of control.

Watch for this pattern in your own systems. Are your probabilistic systems creeping into deterministic territory? The 40% error rate doesn’t announce itself. It hides in the gap between “AI recommends” and “nobody checks anymore.”

While big tech burns billions, you can win with boring

Google is watching $2 trillion in AdWords revenue evaporate but you can bet they’ll figure this out. Most companies are so busy chasing their own AI strategies that they haven’t noticed the ground shifting beneath them.

Who ends up in the 20% who succeed? Not the companies with the biggest models or flashiest demos.

“We’re in the age of execution right now. The data is everywhere. The algorithms are everywhere. It’s a question of applying them to your particular business to get something done.”

— Eric Kavanagh, AI Analyst and host of DM Radio

McKinsey’s consulting knowledge? “Out in the wild,” Kavanagh says. Proprietary algorithms? Commoditized. Kavanagh remembers the dotcom boom, sitting in the Empire State Building asking which way money was flowing. Same chaos, different technology. The companies that survived weren’t the ones with the best ideas—they were the ones that executed.

“The big guys, they are hemorrhaging cash in the hopes of securing this battleground, this land, you know, like it’s Eastern Ukraine or something,” Kavanagh observes. “But by the time it’s all said and done, I think all the buildings are going to be blown up.”

OpenAI, Microsoft, Google, and Anthropic are burning billions on infrastructure wars. Your battleground is different: execution. Tedious problems. Audit logs. Deterministic decisions.

Unglamorous? Absolutely. Effective? That’s what 20 years of watching AI implementations taught Kavanagh.

Look at your current AI initiatives. Which ones pass the Manus test? Which ones confuse audit logs with bureaucracy? Which ones let probabilistic systems make deterministic decisions?

Here’s the uncomfortable truth. You’re already in one of the two groups. You’re building what works, or you’re building what demos well. You’re in the 20% who asked the hard questions before starting, or the 80% who are about to learn expensive lessons.

Eric Kavanagh has watched this movie before. The ending isn’t a mystery.

Listen to the full conversation with Eric Kavanagh on the Data Faces Podcast.

Connect with Eric: info@dmradio.biz | DM Radio on YouTube (500+ episodes)

Based on insights from Eric Kavanagh, AI analyst, syndicated radio host of DM Radio, featured on the Data Faces Podcast.

Podcast Highlights - Key Takeaways from the Conversation

[2:38] The ready-fire-aim problem killing AI projects “We’re in this sort of ready fire aim mode. Many companies are pulling the trigger on programs, and that’s why you’re seeing 80% failure rates, 95% failure rates. People didn’t really think through what they were trying to do with this stuff.”

[5:53] The Manus test: When impressive demos fail reality Kavanagh tested AI agent Manus by asking it to book plane tickets for his wife. “I was wildly faster myself doing things the old fashioned way than Manus was, and Manus came up with a price that was twice as much.” The pattern: AI looks impressive in demos but fails when you actually use it.

[5:53] What AI success actually looks like “It’s personal optimization of your time, of your productivity.” The implementations that work aren’t sexy—they’re doing tedious, focused tasks in the background. Anti-money laundering. Compliance automation. Things where success is obvious and repeatable.

[9:01] AI judges and juries are coming “In a few years, you’ll start to see some front runners do this, you’ll be able to choose AI judge and jury, or real judge and jury. And my recommendation is, if you’re guilty, go with the real judge and jury. If you’re innocent, go with the machines.”

[11:52] Why guardrails are overrated “Guardrails in general are overrated. I think that they’re very difficult to enforce.” The problem: you can’t build perfect guardrails around systems that nobody fully understands, including the people who designed them.

[11:52] The minimal viable governance framework “Audit logs. Just audit logs, as long as you’ve got some log file that says what it did where... You have to be able to kill the pods basically that launch whatever structure you have.” Four questions: What did it do? Where? Can you stop it? Can you remediate?

[16:10] The $2 trillion disruption nobody planned for “About $2 trillion have been paid to Google AdWords in the past 20 years... That’s out the window now.” ChatGPT gives answers directly—organic web traffic is plummeting. “Companies are freaking out. Google’s empire is teetering.”

[17:44] How error rates compound A company chaining multiple LLM calls together found “there’s an error rate at every step of the way, and when you multiply that out, the error winds up being like 40% or something.” Medical AI is getting recommendations wrong 50% of the time.

[19:39] Never mix deterministic with probabilistic systems “Databases aren’t going away. Transactional systems aren’t going away. They will be aided and abetted by these other systems, but you have to be careful not to mix those two.” For most business decisions, you need certainty.

[22:00] Small models will rule the day “I think small language models, or just old fashioned, deterministic AI models, are going to be ruling the day, at least I hope so, because the big stuff is too big, and it’s unwieldy and you can’t trust it.”

[24:27] IP is dead—we’re in the age of execution “IP is kind of dead. Intellectual property is dead. Copyright is dead... We’re in the age of execution right now. The data is everywhere. The algorithms are everywhere. It’s a question of applying them to your particular business to get something done.”

[38:00] Big tech is hemorrhaging money “Open AI, Microsoft, Google, Anthropic, they’re hemorrhaging money because they want engagement. They’re trying to win this battle... but they’re forgetting that we can leave.”

About David Sweenor

David Sweenor is an AI, generative AI, and product marketing expert. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

• Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

• Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

• The Generative AI Practitioner’s Guide: How to Apply LLM Patterns for Enterprise Applications

• The CIO’s Guide to Adopting Generative AI: Five Keys to Success

• Modern B2B Marketing: A Practitioner’s Guide to Marketing Excellence

• The PMM’s Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter@DavidSweenor and connect with him onLinkedIn.

Most AI agent pilots never make it to production

CIOs are burning millions on AI agent pilots that die in the sandbox. This isn’t unexpected; it’s a predictable pattern. Since we’re at maximum hype, executives see flashy demos, demand results, and are looking for the ever-elusive ROI. Data scientists, AI engineers, and IT teams scramble to build something but six months later, the pilot sits unused while the next shiny object captures boardroom attention.

About Catalina Herrera

Catalina Herrera is the Field Chief Data Officer at Dataiku and a Colombian-born electronic engineer with over 25 years in the United States. She holds multiple master’s degrees in computer science, engineering technology, and data science. With 20+ years in advanced analytics, she has worked across different roles and technologies, from hands-on data science projects to enterprise consulting. Catalina helps organizations deploy machine learning and AI use cases that maximize data opportunities. In her spare time, she’s a DJ who uses AI to create her own music, embodying her philosophy of humans empowered by AI.

“A lot of people don’t really understand what agents are and what they bring to the table,” Catalina explains. “That’s going on a lot in the field.”

So what exactly are AI agents? Catalina describes them as systems that combine multiple types of intelligence to act autonomously. Think of an agent as software that can access your descriptive analytics (dashboards and reports), your predictive models (forecasting and risk algorithms), and generative AI capabilities, then orchestrate all these assets to complete tasks without constant human direction.

“The agentic layer happens when you combine all of these techniques that you are applying in terms of the question that you are asking of the data,” she explains. But it’s not just about technology. Success requires what she calls a “multi-variable model” that coordinates “all of those data sets and data outcomes, plus the people, plus the experts, plus the SMEs and everything else that needs to be part of that multi-variable model for this to be successful at the enterprise.”

Many think of AI agents as glorified chatbots, but they can be much more sophisticated than that. Agentic systems can leverage decades of organizational intelligence, human expertise, and institutional processes to “think” and then take autonomous actions without human intervention. But before this can become a reality, you need AI-ready data to provide the appropriate context.^[1]

The technology is evolving at breakneck speed, but your workforce is likely already using agentic capabilities in ChatGPT, Gemini, or Claud. In the race to agentify everything, many organizations approach agents backward, chasing novelty instead of augmenting existing workflows.

“The reality of it is that I personally don’t think that a lot of people really understand what it is and what it brings to the table... And that’s going on a lot in the field. I will say that the higher the level of the way that you think about it, the better.”

— Catalina Herrera, Field Chief Data Officer at Dataiku

The companies that crack this code will create sustainable advantages. The ones that don’t will spend years playing catch-up, wondering why their expensive pilots fizzled out so quickly.

Stop building from scratch. Weaponize what you have.

Consider how this works in practice. In one demonstration scenario, a wind farm operator was drowning in maintenance data. Twenty years of sensor readings. Multiple predictive models for different turbine types. Maintenance crews with decades of expertise. All disconnected.

The agentic breakthrough wasn’t building something new. But rather, they simply connected existing assets through an intelligent layer that could understand and instantly act on institutional knowledge.

In this hypothetical deployment, a maintenance manager could type: “Email my crew the three turbines most likely to fail next week.” The system would pull sensor data, run predictive models, analyze failure probabilities, and send detailed maintenance orders. What previously required hours of manual analysis would happen in seconds.

“Ask the agent in a conversational user interface. Hey, send the email to my maintenance crew and attach to the email the top three turbines they need to focus on next week period... that is ROI right there, how many hours you saved in between, in terms of all the data gathering assets.”

— Catalina Herrera, Field Chief Data Officer at Dataiku

This pattern scales across various industries. For example, financial services may combine transaction monitoring with fraud detection models and regulatory knowledge bases to enhance their capabilities. Retailers can merge sales forecasting with inventory optimization and supplier data to improve their operations. Healthcare organizations often connect patient records with clinical trial protocols and drug interaction databases.

Your organization already has the intelligence. The question is whether you’ll connect it effectively or keep managing data silos manually while competitors automate.

Four questions that separate success from failure

Most failed pilots skip the fundamental evaluation steps. Teams often jump to implementation without ever defining success criteria or operational requirements. Catalina’s approach treats agents like enterprise systems that need a structured assessment.

What are you actually trying to accomplish? Which specific process will improve? Who will use this system? What metric will change? You need an executive sponsor who cares about the outcome, not just the technology.

How will you control the system? What prompts will users write? Which data sources will it access? What approval workflows must it follow? Are there regulatory constraints or policy requirements?

“First of all, you have to classify that use case into the four-part framework, which consists of delegation, description, assignment, and diligence. You need to know, first of all, the what and the why.”

— Catalina Herrera, Field Chief Data Officer at Dataiku

How will you judge performance? What constitutes a hallucination in your context? Which datasets will you use for testing? How will you compare different language models? What’s your acceptable cost and latency threshold?

Who will operate this long-term? Who monitors daily performance? Who approves system changes? How do you collect user feedback and incorporate improvements? What’s your process for handling model drift or data quality issues?

Start with internal use cases before customer-facing deployments. Internal pilots and proven AI deployment strategies let you refine the system and understand failure modes without external risk.^[2]

The hidden risks of chained AI systems

We all know by now that simple ChatGPT queries hallucinate all of the time. What happens when you chain five AI agents together in a multi-step workflow? Does the error rate compound exponentially? I think of it like the butterfly effect – you know, you know, if it flaps its wings in some far-flung region of the world, which leads to a hurricane on the other side.

The concern is legitimate. Multi-agent systems introduce new failure modes that traditional software doesn’t have. Consider a hypothetical scenario where one agent analyzes suspicious transactions, passes recommendations to another agent for regulatory reporting, which triggers a third agent to file compliance documents. When the first agent misclassifies a legitimate transaction, the error cascades through the entire pipeline. Even worse, there can be small perturbations that compound and may go undetected at any single step.

“Keep the human in the loop, is not a joke... what it means to keep the human in the loop with an agentic flow that can be non-deterministic and can hallucinate, and that goes back to your original goal. What is it that you are trying to accomplish?”

— Catalina Herrera, Field Chief Data Officer at Dataiku

Catalina’s solution focuses on strategic human oversight rather than trying to eliminate uncertainty. “Keep the human in the loop is not a joke,” she emphasizes. “What it means to keep the human in the loop with an agentic flow that can be non-deterministic and can hallucinate goes back to your original goal.”

To help reduce uncertainty, system developers need to design appropriate checkpoints and oversight. Different language models produce different outputs for identical inputs, so systematic testing becomes essential. Token consumption can spiral out of control without rate limiting. End-to-end lineage tracking becomes mandatory when decisions are made across multiple systems.

The governance lesson from business intelligence applies directly. Organizations that didn’t manage BI deployments systematically ended up with thousands of conflicting reports and dashboards that nobody trusted. This still happens today – what could possibly go wrong with agents? It’s essential to design for controlled improvement of existing processes, not perfection.

Blockbuster had every advantage but missed the moment

Remember when you used to make it a “Blockbuster night”? They dominated video rental with 9,000 stores and deep customer relationships. They understood the movie business better than any startup. But, as history tells, they misread the inflection point where technology capability met consumer demand.

Netflix didn’t have better movies or superior customer service. They recognized that DVD-by-mail could replace store visits, and then that streaming could replace physical media entirely. Blockbuster saw the technology but missed the competitive timing.

A similar inflection point is happening today with AI agents. Although changing quickly, the technology is production-ready (at least for internal use cases). Your workforce expects intelligent tools. Early adopters are gaining measurable advantages while others debate whether to act.

“This is the opportunity for you to be the Netflix and not the Blockbuster. How are you going to ensure that you are going to maximize the opportunity that this brings for you and for your teams... this is the moment where if you do so, and if you do so right, is going to be a very clear differentiator in terms of your competitive landscape, so the time is now.”

— Catalina Herrera, Field Chief Data Officer at Dataiku

Manufacturing companies are increasingly automating quality control decisions that previously required human experts. Insurance firms are experimenting with processing claims in minutes instead of days. Logistics providers are testing real-time route optimization based on traffic, weather, and delivery constraints.

These represent early production deployments delivering measurable business value while many organizations remain in pilot phases.^[3] “That is ROI right there, how many hours you saved in between, in terms of all the data gathering assets.” But, measuring real AI business value requires more than tracking time saved—it demands understanding actual business outcomes.

The window for competitive advantage remains open but appears to be narrowing. Organizations that establish systematic approaches to AI agent deployment may create sustainable advantages over those still debating implementation strategies.

The competitive landscape suggests that timing matters as much as execution. Whether organizations lead or follow in this transformation will likely depend on decisions made in the coming months rather than years.

Based on insights from Catalina Herrera, field CDO at Dataiku, featured on the Data Faces Podcast.

Podcast Highlights - Key Takeaways from the Conversation

[0:53] What AI agents actually are Catalina explains that agents orchestrate descriptive analytics, predictive models, and generative AI into a multi-variable model that includes “all of those data sets and data outcomes, plus the people, plus the experts, plus the SMEs and everything else that needs to be part of that multi-variable model for this to be successful at the enterprise.”

[6:05] Why most pilots fail “The reality of it is that I personally don’t think that a lot of people really understands what what it is and what it brings to the table. And that’s going on a lot in the field. I will say that higher the level on the way that you think about it, the better.”

[14:38] The four-pillar evaluation framework Catalina introduces delegation (what and why), description (how to instruct), assignment (how to judge), and diligence (how to operate). “You need to know, first of all, the what and the why. What is it that you are trying to accomplish? Who is going to be using these. What is the KPI that you are targeting to move?”

[15:24] The garbage in, garbage out reality “Now you have a layer there that is a very interesting layer, which is, now you are thinking about an AI system so you cannot come from your individuality, as in, I am building this one model and see how that one model is going to perform here. Now you have to think backwards.”

[19:53] Digital interns that leverage existing intelligence Catalina frames agents as “digital interns” that can access organizational knowledge and act on it. The predictive maintenance example shows how agents can combine 20 years of sensor data, multiple predictive models, and domain expertise into a simple request: “Send the email to my maintenance crew and attach to the email the top three turbines they need to focus on next week.”

[24:30] Human-in-the-loop is not optional “Keep the human in the loop, is not a joke. What it means to keep the human in the loop with an agentic flow that can be non deterministic and can hallucinate, and that goes back to your original goal. What is it that you are trying to accomplish?”

[32:24] Cost control and guardrails Catalina emphasizes the importance of rate limits, golden data sets, and monitoring token consumption. “You don’t want surprises. There are a lot of surprises so far in the field in terms of the bill from the tokens now on these llms, so it’s something serious to consider.”

[36:15] Avoiding agent sprawl When asked about the risk of repeating BI mistakes with too many agents, Catalina acknowledges: “Yes, it is the same risk, but I think we have a couple of lessons learned from the previous decade.” The solution is a repeatable framework: “Once you do it right again, you do it right once, and then copy paste.”

[39:56] The competitive differentiation moment “This is the opportunity for you to be the Netflix and not the blockbuster. How are you going to ensure that you are going to maximize the opportunity that this brings for you and for your teams... this is the moment where if you do so, and if you do so right, is going to be a very clear differentiator in terms of your competitive landscape, so the time is now.”

Based on insights from Catalina Herrera, field CDO at Dataiku, featured on the Data Faces Podcast.

^[1] Sweenor, David. “Generative AI’s Force Multiplier: Your Data” TinyTechGuides, October 14, 2023. https://tinytechguides.com/blog/generative-ais-force-multiplier-your-data/

^[2] Sweenor, David. “Generative AI Deployment Strategies: A Strategic Guide for CIOs and CTOs.” TinyTechGuides, May 21, 2024. https://tinytechguides.com/blog/generative-ai-deployment-strategies/.

^[3] Sweenor, David. “How to Build a Compelling Business Case for Generative AI.” TinyTechGuides, September 22, 2024. https://tinytechguides.com/blog/how-to-build-a-compelling-business-case-for-generative-ai/

A marketing leader spent over $1 million on a major industry summit this year. When asked about expected pipeline return, the conversation got uncomfortable fast. This is not uncommon.

Meanwhile, Rajeev Kozhikkattuthodi had his own AI reality check. He used to love using emojis in his business writing until AI overindexed on them and polluted content everywhere.^[1] Now, if you use emojis in your writing, you’ll simply sound like a bot.

The problem in both cases wasn't the technology itself. It rarely is. It was the gap between having great insights and being able to act on them in a timely fashion.

About Rajeev Kozhikkatththodi

Rajeev Kozhikkattuthodi is the co-founder and CEO at Poexis, bringing 24+ years of experience in product engineering, sales, and the "messy bits" of building scalable businesses. He's helping growth marketers figure out how agentic AI can transform events, ABM, and inbound strategies from analysis engines into action drivers. Connect with Rajeev on LinkedIn.

In our conversation, we explore how B2B marketing leaders can move beyond endless data analysis, searching for the perfect insight, to drive measurable outcomes in 30-90 day cycles. You'll learn why most AI projects stall at the pilot stage, how to turn events from awareness plays into pipeline generators, and what leadership skills matter most in an AI-augmented world.

The mindset shift marketers need to make

Rajeev Kozhikkattuthodi has watched marketing teams build incredible analytical capabilities over the past decade. They can segment audiences down to specific individuals. They are able to track attribution across seventeen different touchpoints. They generate insights that would have been impossible just five years ago. The technology has come a long way; you no longer have to be an expert. Simply ask a question of your favorite tool and out comes an analysis and recommendation.

So, what’s the rub? Well, most of them are still missing their numbers.

"The number one mindset shift that marketers and sellers have to look at is how do I move from just analysis to agency," says Kozhikkattuthodi, CEO of Poexis.

The disconnect isn't about data quality or analytical sophistication. Marketing teams have those bases covered.^[2] The problem is execution speed. You identify the perfect prospect segment, but someone still has to research each company manually. You know exactly what content performs best, but creating and atomizing it at scale while adapting it to hyper-specific audiences and companies remains problematic.

The problem lives in the space between insight and action. Picture if you will….a dimension where brilliant insights exist but never quite become reality. You've just crossed over into...the marketing Twilight Zone.

Kozhikkattuthodi calls this "analysis paralysis."^[3] Teams generate insights they can't actually act on fast enough to matter.

Agentic AI changes significantly this dynamic. Instead of just telling you what to do, it does things on its own volition (if you want it to). It researches prospects while you sleep. It personalizes outreach sequences based on real-time triggers. It creates content variations, automatically tests them, and adapts them based on data-driven metrics.

The shift requires letting go of the idea that more analysis equals better results. Sometimes the 80% solution implemented tomorrow beats the perfect solution delivered next quarter. Rajeev recommends "leveling up" gradually, which he describes as a process of starting with low-risk automated actions, measuring results, then expanding autonomy based on what works. Think of it as training wheels for AI adoption. Start with automating research or basic content generation, then progress to more complex decision-making as you build confidence and guardrails. In other words, don’t try to boil the ocean; take baby steps.

Why most AI projects stall (and how to avoid it)

The MIT NANDA study that found 95% of AI projects fail wasn't telling marketing leaders anything they didn't already suspect. Most have a graveyard of pilot programs that showed promise in fancy vendor demos but never scaled. This connects directly to the marketing Twilight Zone problem. Teams can analyze endlessly, but turning insights into measurable business outcomes remains elusive.

The Poexis CEO sees three failure modes when he talks to marketing teams about their AI experiments.

First is the ROI articulation problem.^[4] "The biggest reason most of these pilots around AI are stalling is the ability to look at ROI. The number one reason that most people get wrong is being able to articulate ROI at the P&L level," says Kozhikkattuthodi.

Teams get excited about productivity gains. Content creation is 10x faster. Research that used to take hours or days now takes minutes. But when the CFO asks how this translates to pipeline or revenue, the conversation stalls. Individual productivity doesn't automatically translate into business outcomes.

The second trap is what the Poexis founder calls "spray and pray at scale." AI makes content generation so easy that teams flood their channels. They create 100x more blog posts, social media content, and email campaigns that never make it past the spam filters and are simply ignored. But more content doesn't equal more qualified leads. Without a strategy behind the volume, you're just adding to the noise.

The third failure mode is analysis paralysis at the technology level. New AI models launch weekly. Teams spend months evaluating options instead of implementing solutions. Rather than waiting for perfect tools, Rajeev recommends picking a narrow use case, choosing a good enough technology, and starting within 30 days.

The companies that succeed focus on outcomes first. They identify specific business problems, set measurable goals, and use AI as a tool to achieve those goals rather than as a solution looking for problems.

Practical applications where agentic AI drives results

The martech landscape looks a lot like the patterns on Kozhikkattuthodi's and Sweenor’s Hawaiian shirts that they wore to the Podcast. You could spend five years analyzing every tool and still miss half the players. But while martech leaders debate technology stacks, buyers are changing how they want to be engaged.

"What we're seeing, somewhat ironically, is that buyers are really tuning out of all this digital noise. What they are preferring is really super relevant, personalized, and oftentimes in-person experiences," says Kozhikkattuthodi.

This shift creates opportunities for marketers willing to think beyond traditional automation.

Events: From awareness plays to pipeline generators

Take the marketing leader who spent over $1 million on that industry summit. Most event teams focus on attendance numbers and brand awareness. But Rajeev works with teams that set different goals. They set SMART goals like doubling pipeline output from the next event.

The change starts with pre-event research. Instead of generic outreach to drive attendance, agentic AI pulls intelligence from dozens of sources about target attendees. Not just what's in your CRM, but insights from recent company announcements, hiring patterns, and industry moves.

This research enables what Rajeev calls "experiential relevance." Instead of everyone seeing the same booth demo, each conversation gets tailored to what that specific company is trying to solve. Instead of generic follow-up emails, post-event outreach references actual conversations and next steps.

Here's how this works in practice: A cybersecurity company used AI to research 200 target attendees before a major conference. The system identified that one prospect's company had recently announced a cloud migration initiative. At the booth, the sales rep opened the conversation by asking about migration security challenges rather than delivering a standard product pitch. The prospect was impressed by the relevance and scheduled a follow-up call that same week.

"In the next 90 days, I'm gonna double my pipeline coming out of these events. That's the goal that I'm gonna set for myself my next event, and my team," says Rajeev.

ABM: Research and orchestration at scale

Account-based marketing typically breaks down at the research and activation phase. Sales teams want detailed intelligence on target accounts, but gathering it manually doesn't scale. Agentic AI can research hundreds of accounts simultaneously, identifying decision makers, recent business changes, and engagement opportunities.

The intelligence becomes actionable through orchestration. Instead of batch-and-blast campaigns, you can trigger personalized sequences based on specific account behaviors and characteristics.

Inbound: Staying human in the age of zero-click searches

HubSpot pioneered inbound marketing with the promise that great content would attract qualified leads. But search behavior has shifted. More searches now result in zero clicks as gen AI services and AI agents answer questions directly without sending users to websites.

Rajeev suggests building "agentic experience hubs" rather than traditional blog content. Instead of hoping people find your articles, create content specifically designed to be cited by AI research agents. This means more conversational formats, deeper context, and quantitative backing for claims.

The goal isn't driving traffic to your website. It's becoming the authoritative source that AI agents reference when prospects research your space.

What marketing leaders need to do now

The advice sounds counterintuitive coming from someone who spent decades in engineering and product development. But Kozhikkattuthodi believes the next few years will reward marketing leaders who focus less on technical skills and more on uniquely human capabilities.

"What really matters in the next couple of years is going to be more taste, less the hard skills.^[5] To cultivate taste is very human and is surprisingly uncommon," says Kozhikkattuthodi.

Technical marketing skills have become table stakes. Most marketing teams can run attribution analysis, set up marketing automation, and interpret conversion data. But taste is harder to replicate or automate. It's knowing which campaign concept will resonate before you test it. It's recognizing when personalization feels creepy versus helpful. It's understanding which trends matter and which are just noise.

Cultivating taste requires consuming different perspectives, not just marketing content. Read what your customers read. Attend conferences outside your industry. Talk to people who disagree with your approach. And most importantly, talk to your customers and prospects as much as you can.

The second priority is relationship building. AI can research prospects and generate outreach, but it can't build genuine connections. Focus on strengthening relationships with sales teams, customers, partners, and leadership teams.

These relationships become strategic advantages when everyone has access to similar AI capabilities. Sales teams prioritize leads from marketers they trust. Customers engage with brands that understand their business beyond what's in public databases. Partners collaborate more deeply when relationships extend beyond transactional exchanges.

The third requirement is what Kozhikkattuthodi calls "bias toward action." Instead of spending quarters perfecting strategies, successful marketing leaders will rapidly experiment and iterate based on results.

Set 30-day goals. Pick one agentic AI application and implement it before the month ends. Measure results. Adjust approach. Repeat the cycle.

This approach works because AI technology evolves too quickly for long-term planning to remain relevant. The teams that succeed will be the ones that adapt fastest, not the ones with the most comprehensive strategies.

Escaping the marketing Twilight Zone

The marketing Twilight Zone where brilliant insights exist but never become reality doesn't have to be permanent. Agentic AI offers a way out, but only for teams willing to shift from analysis-first to action-first thinking.

What makes this moment different from previous waves of martech is the speed of change. Traditional approaches like lengthy pilot programs and comprehensive planning can't keep pace. The winners will be marketing leaders who develop what Rajeev calls "taste"—the human judgment to know which applications matter—combined with the courage to implement quickly and iterate based on results.

The irony is striking. In an age where AI can generate infinite content variations and analyze customer behavior at unprecedented scale, the most valuable marketing skills are becoming more human, not more technical. Building genuine relationships, making aesthetic judgments about what resonates, and maintaining bias toward action become competitive advantages precisely because they can't be automated.

Your next 30 days matter more than your next 30-month strategy. Pick one application—events, ABM, or inbound—and start building your "leveling up" muscle. The technology exists. The question is whether you'll use it to escape the analysis twilight zone or remain trapped in it.

Ready to move beyond analysis paralysis? Start with one agentic AI application this month. Pick events, ABM, or inbound. Set a measurable 30-day goal. Take action. Visit Poexis.

Leave a comment

Share B2B Marketing Prompts by TinyTechGuides

Podcast Highlights - Key Takeaways from the Conversation

[2:28] Defining Agentic AI "The crux of the difference that agentic AI brings to the table is the ability to take action, to do something, right? And I think that's really the sort of pivotal point where AI becomes a lot more than what helps you sort of predict, what helps you analyze."

[5:37] The Mental Shift Required "The number one mindset shift that marketeers and sellers and really sort of revenue leaders have to look at this is, how do I move from just analysis to agency."

[16:13] Why AI Projects Fail "The biggest reason most of these pilots around AI are stalling is the ability to look at ROI. The number one reason that most people get wrong is being able to articulate ROI at the P&L level."

[24:09] The Buyer Behavior Shift "What we're seeing, somewhat ironically, is that buyers are really tuning out of all this digital noise. What they are preferring is really super relevant, personalized and oftentimes in person experiences."

[29:01] Staying Human in AI Content "One of the most human ways to communicate, necessarily as copy, but in terms of conversations, is one of the best tips that you can use to drive inbound traffic, especially via agentic sources."

[36:53] Leadership in the AI Era "What really matters in the next couple of years is going to be more taste, less the hard skills. To cultivate taste is very human and is surprisingly uncommon."

Based on insights from Rajeev Kozhikkatthuthodi, co-founder and CEO at Poexis, featured on the Data Faces Podcast.

About David Sweenor

David Sweenor is an AI, generative AI, and product marketing expert. He brings this expertise to the forefront as the founder of TinyTechGuides and host of the Data Faces podcast. A recognized top 25 analytics thought leader and international speaker, David specializes in practical business applications of artificial intelligence and advanced analytics.

Books

● Artificial Intelligence: An Executive Guide to Make AI Work for Your Business

● Generative AI Business Applications: An Executive Guide with Real-Life Examples and Case Studies

● The Generative AI Practitioner's Guide: How to Apply LLM Patterns for Enterprise Applications

● The CIO's Guide to Adopting Generative AI: Five Keys to Success

● Modern B2B Marketing: A Practitioner's Guide to Marketing Excellence

● The PMM's Prompt Playbook: Mastering Generative AI for B2B Marketing Success

With over 25 years of hands-on experience implementing AI and analytics solutions, David has supported organizations including Alation, Alteryx, TIBCO, SAS, IBM, Dell, and Quest. His work spans marketing leadership, analytics implementation, and specialized expertise in AI, machine learning, data science, IoT, and business intelligence.

David holds several patents and consistently delivers insights that bridge technical capabilities with business value.

Follow David on Twitter @DavidSweenor and connect with him on LinkedIn.

^[1] Sweenor, David. "The Great Enshittification of the Written Word." TinyTechGuides.com, August 23, 2025. https://tinytechguides.com/blog/the-great-enshittification-of-the-written-word/.

^[2] Sweenor, David. "Digital Marketing in the Age of AI Disruption Part 2." TinyTechguides.com. June 9, 2023. https://tinytechguides.com/blog/digital-marketing-in-the-age-of-ai-disruption-part-2/.

^[3] Sweenor, David. "Digital Marketing in the Age of AI Disruption Part 1." TinyTechGuides.com, June 2, 2023. https://tinytechguides.com/blog/digital-marketing-in-the-age-of-ai-disruption-part-1/.

^[4] Sweenor, David. "Beyond the AI Hype: What 20% of Companies Get Right." TinyTechGuides.com. February 11, 2025. https://tinytechguides.com/blog/beyond-the-ai-hype-what-20-of-companies-get-right.

^[5] Sweenor, David. "The 'Survival of the Nimblest' Strategy for AI Marketing Success." TinyTechGuides.com, July 1, 2025. https://tinytechguides.com/blog/the-survival-of-the-nimblest-strategy-for-ai-marketing-success/.