Get a team of 15 to 200 people to change how the work is really done — including the two or three who have quietly decided to wait you out.
Most AI programmes in companies past 2M in revenue do not fail on the technology. They fail on a Tuesday morning, quietly, when six people decide it is faster to do it the old way and nobody says so out loud.
You will not see it in the dashboard. Licence counts hold. People mention the tool in stand-ups. There are screenshots in the company channel. Underneath, the actual work — the quote, the handover, the monthly report — comes out exactly as it did in January.
We call that theatre, and it is the normal outcome. Not because your team is lazy or frightened, but because you asked them to change how they work while every incentive you own still rewards the old way, and because two or three people you depend on have decided to wait you out. They are usually right that waiting works. Most programmes are gone in a quarter.
This is the guide to the human half: how to open the programme so people actually move, what to do with the ones who don't — including the ones you cannot afford to lose — and how to read the difference between a team that has adopted and a team performing adoption for your benefit.
No culture workshop. Four conversations, one weekly ritual, one number that cannot be faked.