Harvard Business Review just put a name on the pattern I have been fixing for clients since long before generative AI got a Super Bowl ad.
The article is "When Developing an AI Strategy, Beware the Urgency Trap" by David De Cremer, professor at Northeastern. Read it. It is short. Then come back and I will tell you what an operator sees that a professor sometimes cannot.
"The problem, however, is not that AI does not work. The problem is how leaders think about it." David De Cremer, HBR, July 2026
He is right. The MIT report he cites says 95 percent of generative AI pilots fail. The NBER survey he cites says roughly 90 percent of 6,000 senior executives across four countries report no measurable productivity gain from AI in three years. Those are not soft numbers. Those are executive-desk numbers.
Here is what De Cremer names as the fix: prioritize clarity of purpose, resist urgency bias, focus on advancing the company's vision. Correct in theory. Underspecified in practice.
Let me tell you what actually kills the pilot.
When a CFO says "we need AI in the claims triage workflow this quarter," they are not being intellectually lazy. They are responding to real pressure: the aged claims report, the labor line item, the board question they cannot answer. The urgency is legitimate. The trap is the assumption that any AI project pointed at that urgency will produce a business result.
It will not. I have seen it fail up close, across enterprise payer platform migrations, care management rollouts, claims platform stabilizations, and now inside small businesses using generative tools. The failure mode is almost always the same and almost never named:
The pilot is technically successful and operationally invisible.
The model returns predictions. Accuracy is within acceptable bounds. The dashboard exists. But no downstream workflow was ever restructured to consume the output, no operator was ever trained to trust it, no auditor was ever satisfied it complies, and no P&L line ever moves. Six months later the executive sponsor changes jobs and the pilot dies quietly.
De Cremer says leaders should "focus on advancing the company's vision" instead of chasing urgency. Good sentence, hard to operationalize. Nobody wakes up saying "today I will not advance my company's vision." Everybody thinks they are.
The difference between a pilot that dies and a pilot that scales is not the strength of the vision statement. It is whether the organization has done the diagnostic work to know where it is actually maturing.
That is why I built The Five Pillars of AI Maturity™. Not as another framework in the LinkedIn ether. As the diagnostic instrument I actually use with clients before we touch a model. Five pillars, thirty sub-capabilities, scored on a five-level maturity scale grounded in the Carnegie Mellon capability maturity science and cross-walked against the Six Principles of Responsible AI.
The five pillars are simple to name. They are punishing to score honestly:
Most enterprises score somewhere between Level 1 (ad hoc) and Level 2 (repeatable) across most pillars when they start. They are trying to run a Level 4 (managed and measured) AI pilot on Level 1 infrastructure. It fails. Not because AI does not work. Because the operational floor could not hold the pilot's weight.
The urgency trap has a tell. When the pilot is framed as "we need AI to solve X by Q3," the answer is almost always "you need governance and data hygiene by Q3 more than you need AI." But that is not the answer anyone wants to buy. So they buy the pilot. And the pilot dies.
I have been inside payer operations turnarounds since Y2K. Every major core payer administration platform. Every claims platform conversion. Every clinical care management rollout. I ran claims stabilization at a regional health plan through a full core payer platform conversion. I built the enterprise learning framework at UCare during the financial turnaround. I have watched what actually holds when the pressure comes.
It is not the sexiest deliverable. It is a scored diagnostic, an honest gap plan, and a governance charter you can hand to your board without hedging. That work is what turns a doomed AI pilot into a boring, effective operational upgrade that keeps producing three years later.
De Cremer says leaders should "resist the urgency bias." I say resist the temptation to skip the diagnostic. Same instinct, different exit.
If you have an AI pilot underway and you cannot answer these three questions in one sentence each, you are inside the urgency trap:
If any of those answers start with "we will figure that out later," you have a pilot. You do not have a strategy.
The good news: this is fixable in six weeks, not six quarters. That is what the GGenesis CMM Diagnostic™ does. Six weeks, five pillars, thirty sub-capabilities, scored honestly, with a 90-day activation plan and a 12-month roadmap your CFO can read. Board-ready output. No fluff.
David De Cremer wrote a good article naming a real pattern. Most of what has been published about AI strategy this year is worse than that article. If you are an executive making AI decisions in the next 90 days, read the HBR piece. Then read the Five Pillars methodology. Then decide whether your organization is really ready to move, or whether it needs the diagnostic first.
The urgency you feel is real. The trap is treating it like the strategy.
Gigi Gierbolini-Carrabbia is the founder of GGenesis Strategic Solutions LLC. She spent 25 years running enterprise payer operations before building GGenesis, CaliberSuite™, and Talvori™. Certified in AI Strategies for Business Transformation: Generative and Agentic Intelligence. Based in Broussard, Louisiana. Serves clients nationally.
Describe the pilot you have going, what you thought it would do, and what it is actually doing. We will tell you honestly whether the fix is a CMM Diagnostic or something smaller.
Book a Discovery Call →