TinyTechGuides: Data & AI

The AI slowdown already happened inside your company

David Sweenor — Fri, 25 Sep 2026 18:50:33 GMT

Eric, David, and David on Inside Analysis

On September 18, I joined Eric Kavanagh and David Linthicum on Inside Analysis for an episode Eric dubbed the “AI Slowdown Showdown.” Every major outlet that week was asking whether the frontier AI labs would slow down, and whether they’d do it for our safety or their bottom line. Eric opened the show with fists swinging. “These models are amazing,” he said. “You could talk all day about the amazing things we know they do without having to manufacture these storylines like rogue agents.”

During the first break, I made the point that the slowdown everyone was debating has already happened inside companies, where most generative AI projects stalled long before anyone talked about pumping the brakes. So, if you lead a data or AI program, which slowdown should worry you more? The full episode is below, and here’s how I’d answer that question.

Would an AI slowdown agreement be a cartel?

The debate started with Dario Amodei’s essay “We Must Pace the Frontier,” published on September 12, which called for common safety benchmarks and agreed limits on how fast AI capabilities can grow. Sam Altman, Elon Musk, and Demis Hassabis each voiced their agreement that same day. I read the essay as a repositioning play, since Anthropic already owns the “safe” position in the public’s mind, and I doubt any of it is altruistic in any way, shape, or form. Then I got a laugh out of the panel by pointing out that when three companies restrict output, “I think there’s a word for that. Maybe we’ll call it cartel.”

Eric agreed that it’s “essentially collusion” if the labs are coordinating behind the scenes. Linthicum was less sure about collusion, but he thought the labs were trying to “manipulate the market in their direction. And guess what? It may work.”

I didn’t expect a federal judge to get the question the same day the episode aired. Four paying AI subscribers filed a proposed class action in federal court in California, Buist v. Anthropic PBC, accusing Anthropic, OpenAI, SpaceXAI, and Google of an illegal agreement to slow AI development under Section 1 of the Sherman Act.^[1] The complaint argues that agreeing to progress “slower than competition would otherwise produce” harms paying subscribers.^[2] A complaint is only an allegation, and the frontier might speed up, slow down, or spend the next two years in discovery. Whatever happens, the stalled projects inside your company are really what you need to worry about.

The AI slowdown nobody voted on

When I look at enterprise AI adoption, the brakes went on a while ago. PwC’s 29th Global CEO Survey of 4,454 chief executives in 95 countries and territories found that 56% had seen neither higher revenues nor lower costs from AI, and only 12% reported both.^[3] Those numbers shape the budget conversation for every data and AI leader I know, far more than anything a lab announces.

“I think the slowdown’s already happened. If you look at enterprise adoption, companies aren’t getting value from it. Nobody’s proposed it, nobody’s voted on it, nobody’s signed anything.”
— David Sweenor, Founder/CEO, TinyTechGuides

Linthicum said enterprises are “not doing core applications. They’re not doing inventory control.” They’re using generative AI for productivity work, such as managing a calendar. That work rarely touches the order-to-cash process where a CEO looks for a return, and the split between personal and core use also helps explain how the 3% of companies that win with AI differ from the rest.

On the show, I compared the whole debate to limiting the delivery truck when the warehouse is the bottleneck. Most companies haven’t yet organized the warehouse, which holds the data, processes, and decisions that a model is supposed to plug into. When the shelves are a mess, a faster or slower truck won’t change how many orders go out the door.

What part of your business can be wrong 5% of the time?

Eric asked a question that every CIO should tape to their monitor. Say your model is 80% accurate, you add retrieval-augmented generation (RAG) to reach 90%, and with a lot of work, you push it to 95%. At that point, the question becomes where your business can live with the remaining errors.

“Let’s say it’s 95%. What part of your business can be wrong 5% of the time? Fulfillment, no. Accounting, no.”
— Eric Kavanagh, CEO, The Bloor Group

I offered marketing coupons as a paradigm of hope, and Eric acquiesced on that point. Accuracy is only half the problem with putting a frontier large language model (LLM) into a core business system. Linthicum raised the other half, which is traceability. Core business systems get audited, and an answer that nobody can explain won’t survive a compliance review, however often it happens to be right.

“Every time I’ve tried to look into how I got a bad answer within an LLM, I can’t do it. And guess what? That means I can’t use it for business systems.”
— David Linthicum, Linthicum Research

Hallucination is a defect, and AI agents don’t go rogue

Part of the reason enterprises reach for the wrong fixes is the vocabulary our industry chose for AI. On the show, I asked what would happen if we’d called a hallucination a defect or a bug from the start, since your engineering team would think about it differently. The word “agent” works the same way, because it suggests something with intentions of its own. As I put it on the show, “We’ve given all these humanizing terms, but they’re just computers, computer code doing stuff.”

Eric made the same case about the “rogue agent” stories. In the incidents he described, someone deployed an agent in an environment with open ports, and it followed its objective function with no intent behind it. It didn’t break any rules, because it never decided anything.

“No agent goes rogue. To go rogue, you have to be sentient and decide to go against the rules.”
— Eric Kavanagh, CEO, The Bloor Group

The words matter because each one points a team toward a different fix. A defect gets an owner, a root cause, and a regression test. When teams call the same error a hallucination, the usual response is a shrug and a rewritten prompt, which is why I argued last year that teams should stop chasing hallucinations and focus on agentic quality.

Human in the loop is a process you haven’t designed yet

When Eric asked for final thoughts, I went after the most comfortable phrase in enterprise AI governance. When a team isn’t sure what an AI system will do, the default answer is to put a human in the loop. I called it a cop-out on the show.

“Let’s have a human in the loop. That means we haven’t figured out the process yet. And if you’re making decisions at scale, a human in the loop doesn’t work.”
— David Sweenor, Founder/CEO, TinyTechGuides

At thousands of decisions a day, the person reviewing each one either becomes the bottleneck or starts approving things without reading them. Many of the data and AI leaders I interview on the Data Faces podcast describe a version of this problem, too. Everybody wants guardrails on these probabilistic machines, and the guardrails keep getting narrower until I have to ask why the team didn’t write deterministic code in the first place. That tension showed up in the lessons I took from 27 data leaders in 2025, and nobody has resolved it yet.

Eric described a better way to build these systems. “You have to couch them into a workflow,” he said, calling the model for the one step that needs it “and then funnel that back into your deterministic workflow.” Linthicum went further for core systems.

“The core model is going to be hosted by us. We’re going to run it. We’ll build it using our data. It’s going to have a very narrow focus.”
— David Linthicum, Linthicum Research

Both describe a designed process that settles three things before anything goes live. It identifies the single step where a probabilistic model earns its place and wraps that step in deterministic checks that catch out-of-range outputs. It also logs each model call in enough detail to trace a bad answer back to its source. With those in place, the person in the loop has a specific decision to make, at a specific point, with the evidence in front of them.

Where to start with enterprise AI adoption

If your AI program feels stalled, designing the process will move it faster than waiting for a better model. Here are three places I’d start.

Name the decision behind every human in the loop: Wherever a person sits in the loop in your AI plans, write down the specific decision they make and the information they need to make it. When you can’t, designing that process is your first project.
Mark the one probabilistic step in each workflow: Identify where the model does work that ordinary code can’t, and put a deterministic check on either side of it.
Log model errors as defects: Give every bad output an owner, a root cause, and a test, and you’ll build the audit trail your compliance team will ask for.

Fix the warehouse first

When Eric asked why the AI companies are behaving the way they are, I went back to a business school professor who liked to say it always comes down to the fundamental economics of the situation. The frontier labs will speed up or slow down based on what their investors tell them, and none of us gets a vote on that. What happens inside your own company is a different story, because the warehouse belongs to you, and so does the decision to fix it first.

If you want to work through these ideas in person, join Eric, me, and a room full of practitioners at the Agentic Road Show on Thursday, October 22, at the University of Pittsburgh. Dr. Mark Brady, former chief data officer of the U.S. Space Force, will present his six laws for safe AI, and Carnegie Mellon’s Philip Koopman will close the day on autonomous vehicle safety. Workflow design is most of what I do in my advisory work, and it’s where I see stalled programs start moving again, so I invite you to join us in Pittsburgh.

Frequently asked questions

Is enterprise AI adoption slowing down?

In most companies, enterprise AI adoption slowed well before the frontier labs talked about slowing down. PwC’s 29th Global CEO Survey, published in January 2026, found that 56% of 4,454 chief executives had seen neither higher revenues nor lower costs from AI, and only 12% reported both. Most enterprise use of generative AI sits in personal productivity tasks, such as email and calendars. It has not reached core systems like inventory control or order entry, which is where CEOs look for a return.

Are AI companies being sued over the AI slowdown?

Yes. On September 18, 2026, four paying subscribers to ChatGPT, Claude, Grok, and Gemini filed a proposed class action, Buist v. Anthropic PBC, in the U.S. District Court for the Northern District of California. The complaint accuses Anthropic, OpenAI, SpaceXAI, and Google of an illegal agreement to slow AI development in violation of Section 1 of the Sherman Act, citing executives’ public agreement with Dario Amodei’s September 12 essay, “We Must Pace the Frontier.” The plaintiffs argue that coordinated limits on AI progress reduce the value of paid subscriptions, and the claims have not been tested in court.

What does human in the loop mean in AI, and why isn’t it enough?

Human in the loop means a person reviews or approves an AI system’s output before it takes effect. It becomes a problem when teams use it in place of process design, because it usually means nobody has decided what the person is checking or why. When a system makes thousands of decisions a day, the reviewer either becomes a bottleneck or approves outputs without reading them. A designed process names the specific decision the person makes, at a specific point, with the evidence in front of them.

Is an AI hallucination a bug?

For enterprise purposes, treat an AI hallucination as a defect. Calling a wrong output a hallucination makes it sound like an unavoidable quirk of the model, and teams tend to respond by rewording the prompt. Treating it as a defect gives the error an owner, a root cause, and a regression test, the same as any other software bug. That discipline also builds the audit trail that compliance teams need for AI systems that get reviewed.

Can AI agents go rogue?

AI agents don’t go rogue in the sense of deciding to break the rules, because that would require intent. As Eric Kavanagh put it on DM Radio’s Inside Analysis, “To go rogue, you have to be sentient and decide to go against the rules.” The incidents reported as rogue AI agents usually involve an agent deployed in an environment with open ports or weak containment, where it pursued its objective function as designed. Those incidents call for a security review of the deployment environment.

How accurate does AI need to be for enterprise workflows?

The required accuracy depends on what a wrong answer costs. Even at 95% accuracy, a large language model (LLM) is wrong one time in 20, which is unacceptable for fulfillment, accounting, and most other core processes, though tolerable for low-stakes tasks like marketing coupons. Enterprise AI systems also need traceability, since an answer that nobody can explain won’t survive an audit. Many teams handle both problems by calling a model for one narrow step and wrapping it in deterministic checks and logging.

Where should enterprise leaders start with a stalled AI program?

Start by designing the process around the model before waiting for a better model. Write down the specific decision behind every human in the loop in your AI plans, and mark the single step in each workflow where a probabilistic model does work that ordinary code can’t, with deterministic checks on either side of it. Then log model errors as defects, with an owner and a root cause, so the AI program builds an audit trail as it goes.

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.

Footnotes

^[1]Quinn Wilson. “OpenAI, Anthropic, Google, SpaceXAI Hit With Antitrust Lawsuit.” Bloomberg Law, September 18, 2026. https://news.bloomberglaw.com/litigation/openai-anthropic-google-spacexai-hit-with-antitrust-lawsuit. Case: Buist v. Anthropic PBC, No. 3:26-cv-10693 (N.D. Cal.).

^[2]Kaitlyn Huamani. “Lawsuit says Anthropic, OpenAI, SpaceXAI and Google made illegal agreement on AI slowdown.” Associated Press, via OPB, September 20, 2026. https://www.opb.org/article/2026/09/20/lawsuit-says-anthropic-openai-spacexai-and-google-made-illegal-agreement-on-ai-slowdown/.

^[3]PwC. “PwC’s 29th Global CEO Survey: Leading Through Uncertainty in the Age of AI.” PwC, January 19, 2026. https://www.pwc.com/gx/en/issues/c-suite-insights/ceo-survey.html.

AI's 3 Ws are killing enterprise ROI

David Sweenor — Fri, 24 Apr 2026 13:30:59 GMT

Gartner Orlando 2026 - Photo by Author David E Sweenor

Four out of five organizations are deploying AI. Only one in five is hitting its ROI targets.^[1] I’ve been to the Gartner Data & Analytics Summit eight or nine times over the years, and I’m not sure if it’s déjà vu or a movie I’ve already seen.

When big data was en vogue, we saw the same thing. Everyone rushed to adopt the technology, even though few understood it or could articulate the returns. The pay was always “next year.” Big data had its 3 Vs of Volume, Velocity, and Variety. Some people stretched it to 7, because apparently three wasn’t enough to describe the problem. The AI era already has its own shorthand, and I’m calling them the 3 Ws – washing, wishing, and waiting.

I spent three days at the summit in Orlando this March, and the 3 Ws were everywhere. They explain why that 1-in-5 number is so low, and they’re the same symptoms that dragged big data down a decade ago.^[2]

Washing

The vendor side and the practitioner side might as well have been at two different conferences. On the expo floor, every booth was a native “AI agent” company. I couldn’t figure out what half of them did because the messaging was identical. Big data washing became AI washing, which became agent washing, and the label keeps rotating while the behavior underneath stays the same.

A Gartner analyst warned against this pattern in a session on AI agents in analytics and BI, recommending evidence-based evaluation with real-world scenarios instead of trusting vendor marketing claims.^[3] Good advice, but the expo floor didn’t get the memo. Booth copy sounded like it was written by the same LLM, and vendor presentations felt forced. The AI slop is feeding itself, and Gartner predicts a $58 billion market shakeup as GenAI and AI agents challenge mainstream productivity tools.^[4]

“Big data washing became AI washing, which became agent washing.”
— David Sweenor, Founder/CEO, TinyTechGuides

Wishing

The hallway conversations told a different story compared to the keynotes. Practitioners weren’t talking about what’s working, but about what isn’t working yet. Session after session pointed to 2027 and 2030 as the years when AI will finally achieve nirvana and deliver on its promises.^[5] The finish line keeps moving, and big data had the same timeline problem. The returns were always coming “next year,” just around the corner, right after you got your data sh*t in order.

Six out of ten IT leaders are worried about AI agent cost overruns, but only two out of ten D&A leaders share that concern.^[6] That disconnect should worry everyone. The people closest to the data aren’t worried about the costs, and the people paying the bills are.

“The returns were always coming next year, just around the corner.”
— David Sweenor, Founder/CEO, TinyTechGuides

We might be one bad quarter away from CFOs redirecting AI budgets to other mission-critical priorities, or worse yet, replacing agents with actual employees. When results don’t materialize, the panacea that is AI dissipates, which is what happened to big data. The enthusiasm evaporated the moment CFOs started asking tough questions about returns, and while the technology survived, most of the budgets didn’t.^[7]

Share this with your buddy, they’ll thank you.

Waiting

The “vendor conference” sold AI-first futures, while the practitioners who need to make all of this work understood that the foundations aren’t there. Everyone is waiting, and the waiting comes in three flavors.

“Organizations are building AI on top of foundations they don’t trust.”
— David Sweenor, Founder/CEO, TinyTechGuides

Waiting on the data

Most catalog and context layer vendors are still built for structured, tabular data. Today’s AI strategies depend on documents, images, and code, which current tooling wasn’t designed to handle, and no one seems to be addressing. Meanwhile, the conversation has moved to metadata, semantic layers, and context layers as the connective tissue that makes AI work. As Scott Taylor puts it, context is the new oil.^[8] By 2027, Gartner expects 40% of IT spending on data management to target multistructured data, and AI data readiness spending will increase sevenfold from 2025 to 2029.^[9] The money is apparently coming. Who’ll build the plumbing? Agents or people?

Waiting on governance

Everyone is pro-governance, and for most companies in the US, that’s about as far as they’ll go. AI governance, the kind that covers model risk management, bias testing, usage controls, and agentic oversight, is barely on the radar.^[10] Only 14% of IT leaders said they were confident their data and content assets are secured and governed.^[11] Organizations are building AI on top of foundations they don’t trust.

Waiting on people

Gartner reports that only 6% of D&A and AI leaders consider themselves fully AI-ready when it comes to people, skills, and change management.^[12] I’m not sure I agree with that number. People are using AI all over the place, whether their organizations are ready or not. The formal readiness programs haven’t caught up yet, but informal adoption has already happened. Change management alone can require up to twice the effort of the implementation itself.^[13] Scott Brinker calls this pattern Martec’s Law, where technology changes exponentially while organizations change logarithmically.^[14] That’s why most companies are still staring at the empty shelf long after the technology has moved three aisles over.

Big data had all three of these problems. Organizations bought the tools before fixing the plumbing, skipped governance, and underinvested in people. The AI era is repeating the same mistakes.

Same plot, different city

The Hangover Part II was essentially the same movie in a different city, with a new hotel and another bachelor party leading to the same bad decisions. That’s what this feels like.

The reasons four out of five organizations aren’t hitting their AI ROI targets are the same reasons big data underdelivered. Only the buzzword changed. The organizations that closed the distance last time invested in the boring stuff, the blocking and tackling.^[15] They focused on governance before it was fashionable, fixed data quality before the next platform purchase, and developed people instead of cutting headcount. None of it was exciting, but all of it worked.

The technology works, and the big data technology worked too. What didn’t deliver was everything around it,^[16] which is where the washing, wishing, and waiting did their damage. No new generation of AI agents will fix that for you.

Frequently asked questions

What are the 3 Ws of AI?

The 3 Ws of AI are Washing, Wishing, and Waiting. Washing is vendors rebranding existing products as “AI agents” without real AI underneath. Wishing is that practitioners are deferring AI ROI to 2027 and 2030, while only one in five organizations hits current ROI targets. Waiting is the shortfall in data readiness, governance, and change management. Together, the 3 Ws explain why most AI investments fall short.

What is AI washing?

AI washing is the practice of rebranding existing products as “AI” or “AI agent” without meaningful AI underneath. At Gartner D&A Summit 2026, nearly every expo booth marketed “AI agent” products with near-identical messaging. The pattern mirrors big data washing from a decade ago. The label keeps rotating while the behavior underneath stays the same.

Why are so few AI deployments hitting their ROI targets?

The 2026 Gartner CIO Survey shows that four in five organizations are deploying AI, but only one in five hits ROI targets. The failure pattern repeats big data’s mistakes. Organizations bought tools before fixing data foundations, skipped governance (only 14% of IT leaders feel confident), and underinvested in people. The technology works; the execution around it keeps failing.

Why are IT and D&A leaders divided on AI cost concerns?

At the 2026 Gartner D&A opening keynote, six out of ten IT leaders said they were worried about AI agent cost overruns, but only two out of ten D&A leaders shared that concern. The people closest to the data aren’t worried about costs, and the people paying the bills are. AI budgets may come under CFO scrutiny before D&A teams expect it.

Where should D&A leaders focus to improve AI ROI?

Focus on the three foundations most organizations are waiting on: data readiness, governance, and people. Gartner expects 40% of IT spending on data management to target multistructured data by 2027, with AI data readiness spending growing sevenfold from 2025 to 2029. Only 14% of IT leaders are confident in governance, and only 6% of D&A leaders are AI-ready on people and change management.

About David Sweenor

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]2026 Gartner CIO Survey, n=2,437.

^[2]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/.

^[3]Macari, Edgar. “Activating AI Agents in Analytics and BI Platforms.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[4]Sallam, Rita. “Top Data and Analytics Predictions for 2026.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL. See also Gartner, “Gartner Announces Top Predictions for Data and Analytics in 2026,” press release, March 11, 2026, https://www.gartner.com/en/newsroom/press-releases/2026-03-11-gartner-afromnnounces-top-predictions-for-data-and-analytics-in-2026.

^[5]Sallam, Rita. “Top Data and Analytics Predictions for 2026.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL. See also Gartner, “Gartner Announces Top Predictions for Data and Analytics in 2026,” press release, March 11, 2026, https://www.gartner.com/en/newsroom/press-releases/2026-03-11-gartner-announces-top-predictions-for-data-and-analytics-in-2026.

^[6]Ronthal, Adam, and Georgia O’Callaghan. “Opening Keynote: Navigate AI on Your Data & Analytics Journey to Value.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[7]Sweenor, David. “The AI-Powered CFO: Why Finance Must Shift from Control to Cognition.” TinyTechGuides, March 11, 2025. https://tinytechguides.com/blog/the-ai-powered-cfo-why-finance-must-shift-from-control-to-cognition/.

^[8]Sweenor, David. “Truth Before Meaning — The Three-Word Fix for Data Management.” TinyTechGuides, April 7, 2026. https://tinytechguides.com/blog/truth-before-meaning-the-three-word-fix-for-data-management/.

^[9]Showell, Nina. “Unstructured Data Is the Missing Ingredient to Prepare AI-Ready Data.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[10]Sweenor, David. “Why Bad AI Governance Kills 95% of Enterprise Projects Before Production.” TinyTechGuides, September 9, 2025. https://tinytechguides.com/blog/why-bad-ai-governance-kills-95-percent-enterprise-projects/.

^[11]2025 Gartner GenAI Enterprise Survey, n=360.

^[12]2025 Gartner Survey, n=353.

^[13]Ronthal, Adam, and Georgia O’Callaghan. “Opening Keynote: Navigate AI on Your Data & Analytics Journey to Value.” Gartner Data & Analytics Summit, March 9-11, 2026, Orlando, FL.

^[14]Brinker, Scott. “Martec’s Law: Technology Changes Exponentially, Organizations Change Logarithmically.” Chiefmartec, June 13, 2013. https://chiefmartec.com/2013/06/martecs-law-technology-changes-exponentially-organizations-change-logarithmically/.

^[15]Sweenor, David. “Why 80% of AI Projects Fail (And the Three Boring Decisions That Save the Other 20%).” TinyTechGuides, October 21, 2025. https://tinytechguides.com/blog/why-80-of-ai-projects-fail-and-the-three-boring-decisions-that-save-the-other-20/.

^[16]Sweenor, David. “The $40B Reason Enterprise AI Projects Fail: It’s Not the Tech.” TinyTechGuides, September 6, 2025. https://tinytechguides.com/blog/the-40b-reason-enterprise-ai-projects-fail-its-not-the-tech/.

The $40B Reason Enterprise AI Projects Fail: It's Not the Tech

David Sweenor — Sat, 13 Sep 2025 12:42:23 GMT

Just as people need a place to relax, they also need a place to experiment. Photo by author David E. Sweenor

MIT NANDA's latest research delivers a brutal reality check. Despite $30-40 billion in enterprise AI investment, 95% of organizations failed to move their initiatives from pilot to production. Dually noted, 90% of employees quietly use personal AI tools for work tasks while only 40% of companies provide official subscriptions.^[1]

Your AI strategy isn't failing because of technology. Tech is rarely the problem. You're solving the wrong problem.

During my decade in ASIC yield characterization at IBM, I watched this pattern play out over and over again. Some teams obsessed with discovering the perfect process recipe before production consistently got outpaced by those who shipped early silicon at 5% yields, characterized failures, and iterated. My job was to find the signal in the noise. One of the many lessons I took away from this experience is that experimentation and iteration are the ultimate keys to progress.

Enterprise AI adoption follows similar patterns. After analyzing hundreds of AI initiatives across several companies, interacting with many clients, 200+ articles, 20+ podcasts, and 10+ books on the subject, the successful minority share one trait. They learned from small, incremental experiments instead of engineering away all risk.

The "AI-ready" fallacy creates the problems it claims to solve

Have you heard the term “AI-ready”? Or is your data “AI-ready”? Well, you need to have both, and I’m not underselling their importance, but it’s not what you think. Having a monolithic AI-readiness program isn’t the answer. Corporate AI-readiness programs indefinitely delay value by ignoring how learning actually works. Organizations pursue comprehensive data readiness initiatives, while employees solve real problems with $20/month ChatGPT subscriptions. And they’re doing it without the data being perfect – go figure. It’s not pristine and never will be.

This represents strategic blindness.

Gartner confirms that AI-ready data "is not something you can build once and for all nor that you can build ahead of time."^[2] Yet CIOs and tech leaders spend months in vendor evaluation cycles while their teams discover what works through daily experimentation.

Shane Murray, Field CTO at Monte Carlo Data, sees the pattern clearly. "The teams making the most progress deploy prototype and production AI products, learn where it breaks, learn where it's biased."^[3] Risk management through controlled experimentation beats comprehensive preparation.

Smart organizations apply right-sized oversight. AI for credit decisions needs high governance rigor. AI for marketing copy and coupons probably doesn't need to be that stringent. Scale governance oversight appropriate to the use case rather than treating every initiative or project like you’re launching nuclear missiles.^[4]

I know someone who would like this, I’d better share.

Your employees already cracked the code

While enterprise initiatives wallow in planning phases, your workforce has built a functioning innovation ecosystem using consumer tools. This isn't a rebellion. They have a job to do, and they’re smart. Plus, it's also market validation.

That 90% employee adoption rate represents massive, unsolicited research happening within your organization. Domain experts became power users regardless of official programs, discovering through trial and error what delivers actual value.^[5]

Personal AI experimentation reveals what enterprise demos miss. Immediate iteration is preferred compared to cumbersome corporate processes and approval workflows. Learning through use beats comprehensive training. Leaders need to enable low-friction testing of ideas without career consequences.

Consumer tools work because they embrace experimental adoption. False starts and course corrections teach more than standing from afar and watching..

Forward-thinking leaders recognize shadow AI usage as a source of competitive intelligence, not a compliance risk. Give them a place to safely experiment, and your employees will solve the adoption problem you're spending millions to crack.

People are your real AI strategy

NewVantage Partners research shows 92% of executives identify cultural change as the biggest impediment to data-driven transformation.^[6] This statistic hasn't budged in years because organizations keep attacking symptoms instead of causes.

Dr. Danny Stout from EY's Intelligence Layer puts it bluntly. Teams must align beforehand, or "there's no way that whatever model you choose is going to be successful."^[7] Robert Lake, who advises companies on AI strategy, observes that business leaders often "paper over their business problems and hope that AI will fix them magically."^[8]

Culture beats technology.

Three elements separate successful AI cultures from the 95% failure rate: 1) safe experimentation, 2) aligned incentives, and 3) internal talent development.

Safe experimentation spaces formalize what your employees have already created behind IT’s back. Teams need risk-mitigated freedom to test ideas without threatening their job security.

Aligned incentives fix the fundamental problem where productivity tools become productivity threats. When finding efficiency might eliminate roles, adoption stalls regardless of the technology’s quality.

Internal talent development leverages existing domain knowledge and expertise. People who understand your customers and problems create more value than external "AI specialists" when given basic AI literacy.

Stop investing primarily in platforms. Empowered, incentivized employees who can experiment safely turn grassroots AI usage into a sustainable advantage.

How to join the successful 5%

Shawn Rogers' BARC research shows only 20% of companies achieve AI maturity benchmarks.^[9] These leaders align AI outputs with business KPIs rather than chasing technology trends. The gap between intention and execution separates winners from the 95% failure statistics.

Based on patterns I've documented across hundreds of successful implementations, here's what works.

Map existing AI usage immediately. Survey what tools teams actually use and problems they solve. Look for results, not compliance violations. This reveals where innovation exists and what business value looks like in practice.

Enable rather than restrict. Build governance frameworks that facilitate experimentation, not prevent it. Create sandbox environments where failure has a limited downside but learning has a high upside. Most corporate AI policies optimize for preventing mistakes rather than maximizing learning velocity.

Start with known problems using available data. Launch focused projects addressing real business pain points with existing information. Deploy, measure, iterate. This matches the exact approach your shadow AI users figured out.

Scale what works. Successful companies "amplify what already works within your organization" rather than pursuing wholesale reinvention.^[10] Connect AI initiatives to existing operational strengths and measurable outcomes.

Enterprise AI failure isn't about technological complexity. It's organizational. Your employees already demonstrated that rapid iteration beats comprehensive preparation. The successful minority learned from them instead of fighting them.

Join the 5% who figured it out.

About David Sweenor

David Sweenor brings 25+ years of hands-on experience implementing AI and analytics solutions across Fortune 500 organizations, including IBM, SAS, Dell, TIBCO, and Alteryx. During his 11-year tenure at IBM, he specialized in yield engineering and predictive analytics, developing systems to optimize semiconductor manufacturing and identify yield loss patterns—experience that revealed how iterative improvement outperforms perfection-first strategies. He has authored six books on AI and generative AI applications, written 200+ articles analyzing AI adoption patterns, and co-developed four patents in semiconductors and SaaS. His marketing leadership has generated $350M+ in attributed pipeline and achieved Gartner Magic Quadrant Leader rankings at multiple companies. David hosts the Data Faces podcast and is recognized as a top 25 AI & analytics thought leader.