Most shops I talk to are in group two.
Spreading the word...
Most shops I talk to are in group two.
One: a real innovation or IT bench already poking at AI. Two: no official team, but a few people in different departments already tried ChatGPT. Three: not looking. Product and process as they are.
Group two is most of small and mid-size. Someone asked a model a question or made an image. That person is already the evangelist, whether the title exists or not. The job is to get the rest of the building to try something useful without a lecture.
Talk like a person. Bring a number.
Skip the jargon. Say what it does for the week, not how the transformer works. A plain-language summary of a technical mess is a fine first demo. Then a real problem: the support queue, the forecast, the report nobody finishes. Cost and a return, even a rough one. A chatbot that eats the repeat tickets is a better slide than "we should invest in AI."
Let people try and fail in public. A little budget and time. Share the ones that worked. Use a model to list ideas, then kill most of them. Train at the level people actually are. You go first. Show the thing you built this month, including the miss. Then hand someone else the next one.