#pilots · 7 essays
AI Pilots — Why They Stall, and What Ships
The enterprise does not have an AI pilot shortage. It has a production shortage: pilots that impress everyone in the demo and then spend eighteen months not shipping. These essays are about that gap — how it forms, what it costs, and the uncomfortable fact that killing pilots early and on purpose is how the winners fund the ones that ship.
If you are new here, start with Why 95% of AI Pilots Fail, then Pilot Purgatory for the escape route. And if you have one specific pilot in mind, I built a free triage that runs it through fifteen questions and returns the verdict a steering committee should be asking for: ship it, fix it, or kill it.
Your Documents Are Not AI-Ready — and Retrieval Will Not Fix That
Twenty years of documentation is inventory, not a corpus. Why retrieval blends contradictions a human would catch — and the ten-question test that proves it.
Your Business Unit Became an AI Pilot Factory — How to Govern It Back Into Production
Monthly workshops, showcase meetings, thirty agents that are really system prompts, nothing live. How a business unit governs its own AI work into production.
Kill More AI Pilots. Seriously.
Gartner says 40% of agentic AI projects will be canceled by 2027. Good. Killing pilots early and on purpose is how winners fund what ships.
Everyone Has AI Agents. Almost Nobody Has Them in Production.
Most enterprises have adopted AI agents. Few run one in production — and Gartner expects 40% of agentic projects canceled by 2027. Mind the gap.
Pilot Purgatory — Why Your AI Demos Never Ship, and How to Escape
Your AI demos impress everyone and ship nothing. Why pilots stall between demo and production — and the triage that gets one live in 90 days.
5 Mistakes Enterprises Make With AI — the Same Five, Year After Year
Five mistakes that show up in enterprise AI portfolios again and again — unowned pilots, vendor-written strategy, governance procrastination — and the fix for each.
Why 95% of AI Pilots Fail — and What the 5% Do Differently
An MIT NANDA report found 95% of enterprise GenAI pilots deliver zero P&L impact. The failure is organizational, not technical — here's what the other 5% do.