Change management is the real AI consultancy deliverable

An AI consultancy engagement is easy to scope around the technical deliverable: the workflow, the integration, the dashboard. It's much harder, and much more important, to scope around the actual determinant of success: whether the people whose job changes because of this system actually use it the way it was designed to be used.

Every AI project I've seen fail after a technically successful build failed for the same reason: nobody on the team that has to use the new system daily was involved early enough to feel ownership over it, so it got treated as something imposed on them rather than something built with them.

The practical fix is unglamorous and rarely billed for separately: involve the actual end users in testing before launch, not after, run a short training session that addresses "what do I do when this is wrong" specifically, and check back at two weeks and two months, not just at launch, because adoption erodes quietly if nobody's watching for it.

The technical build is maybe 40% of what makes an AI project actually stick. The other 60% is change management that most engagements don't scope, budget, or bill for, which is exactly why so many technically sound projects quietly stop being used.