AI consulting & coaching · Zürich
Most teams don't need more AI. They need to know where it actually helps.
I'm Luca Widmer. I work with teams to find the two or three places AI earns its keep, build one of them properly, and coach the people who will run it after I'm gone.
Two ways in
One of these fixes the system. The other fixes who runs it.
Find the thing that works, then build it
For teams who have a budget, a hunch and no shared answer. We narrow the field down to one use case worth doing and get it in front of real users before the quarter ends.
- Use-case audit with effort and payoff attached to each option
- A working system in production, not a slide deck
- Evaluation harness so you can tell if it degrades
- Straight answers about what your data will not support
Make it stick after I leave
For the person who has just been handed AI on top of their day job. Regular sessions, real decisions from your own backlog, and someone to check your reasoning before you commit to it.
- Fortnightly one-to-ones with the person who owns delivery
- Team workshops on prompting, evaluation and review
- Second opinion on vendor pitches and build-versus-buy calls
- A written policy your legal team will sign off
How an engagement runs
Seven weeks from first conversation to something your team owns.
Mapping
Two days of interviews across the teams that would actually use the thing. I leave with a list of candidate use cases, ranked by effort against payoff.
Cutting
We kill most of them. What survives has a named owner, a number it has to move, and an agreed way to tell whether it worked.
Building
One working thing, in production, used by real people. Your engineers build it with me so the knowledge stays in the building.
Handover
A runbook, an evaluation set, and a standing hour every fortnight while your team takes it over completely.
Bring me the problem you think AI might solve.
Half the time I'll tell you it won't, and why. That conversation is free and usually saves more than it costs.