Trinny London AI Programme
The training was day one. Day two was where you found out whether any of it stuck.
The question I was chasing
What can a coach actually change in two days that survives contact with the next Monday?
I was one of the coaches on Trinny London's two-day AI programme, delivered by Lichen AI. I did not design it and I did not run it. I had two of the 25 teams, creative and marketing, and I sat with the creative team for both days. That is the honest size of the job, and it is also what made the question worth asking. Two days can teach a tool. Making it stick has to come from something else in the room.
The constraints
The whole company stopped work: 150 people, two days, the second day ending in pitches to leadership. Whatever the teams walked out with had to be worth two days of not trading.
Neither of my teams had an engineer on it. Day one was fundamentals, day two was build and pitch. Anything we started had to fit inside a day and be legible in a few minutes.
The harder constraint was that these were not beginners. Both teams had used AI for months, were quietly disappointed by it, and had half decided the tools were overrated. That is a worse starting position than never having tried: nothing to fill in, only a settled habit to undo.
The decisions that mattered
Repair the workflow they had before adding anything new. The creative team's prompts were not the problem, their inputs were: no brand references, no examples of good, no statement of what wrong looked like. Another tool would only have given them another place to get average results.
Build for themselves, not for the customer. Both teams pointed day two at their own work, the honest choice at hackathon scale. The team knows the problem, can judge the output itself, and can start using it on Monday without a budget line. A customer-facing concept would have made a nicer pitch and a slower thing to prove.
Scope to what can be shown. The day two pitch decided the size of every idea, and anything we could only describe was too big.
Built with: Gemini, Claude, Midjourney
Where it landed
Both teams shipped something, both pitched it, and both left with a tool aimed at their own work rather than at a slide.
All 25 teams had to build something and put it in front of the people who can fund it. Most AI training ends when the training ends. This one ended with 25 concepts written by the people doing the work, which is a better input to a strategy than a completion rate.
Whether either of my teams is still using what it built is not something I can see from here. Two days was never going to answer that.
Part of the Rolling Waves work archive.