Most AI pilots stall because they were built against the documented process, not the real one. When they meet real data, exceptions and handoffs, costs rise, value is hard to show and nobody can say who owns the risk. Gartner expects over 40% of agentic AI projects to be cancelled by the end of 2027.
George McKenna, Chairman and CTO, ETT Group. Updated 15 October 2026.
The pattern I see most often has four parts.
- The wrong process. The pilot was aimed at a process that looked simple on a whiteboard. The real version has photo invoices, renamed cost centres and exceptions handled from memory.
- No baseline. Nobody measured what the process cost before, so nobody can prove what the pilot saved.
- No owner. The pilot belonged to an innovation team, not to the person who runs the process on a Monday morning.
- No evidence. Risk and compliance couldn't see what the system would do, so they couldn't approve it for production.
Gartner's forecast, published in June 2025, gives the reasons as escalating costs, unclear business value and inadequate risk controls. All three trace back to not knowing how the process really runs.
The fix is to do it the other way round: find out what's really happening (process mining plus time with the people who do the work), agree the number, name the owner, then build.
Source: Gartner, June 2025
Show Your Working
A fortnightly piece on building and buying AI that pays its way. What I'm building, what it costs, and the questions to ask before you sign anything.