Workshops
The question missing from 139 AI courses
4 min read · 4 August 2026
This September the Swiss {ai} Weeks run across the country, 139 events in one programme. We read every single one, because we wanted to offer an event ourselves and had to know what already exists. Around forty explain what AI is. A good dozen show how to build with it. And not one asks the question that comes after the building.
Two years ago, “what is AI” was the right first question, and the courses for it were needed. By now, any team can produce something in an hour that looks finished: a concept or a whole campaign. Producing has become cheap. So the problem moves one step along, to the question of whether what got produced is any good. And nobody in the programme asks it.
Why this question, of all things, is missing
It’s the most uncomfortable one. A beginner course ends on a sense of achievement, because something is on the screen at the end. Whether it’s good only shows later, with the customer or in the boardroom. Add to that a property of machine work no beginner course covers: the intern’s text with mistakes gives itself away, clumsy and full of typos. The machine’s text with mistakes arrives in flawless, fluent prose. How we handle that in our own work is in the post about the 565 checked websites.
Bad human work looks bad. Bad machine work looks good.
How to spot a course worth taking
Three things you can check before signing up. Does the course work on your own material or on prepared examples? Is something running at the end that you’ll use in the business on Monday? And is there an answer to the question of how you’ll tell, afterwards, whether a result is good enough to let out, and who decides that?
The third question is the fastest test. Ask the provider before you book. If a concrete answer comes back, you’ve found a good course. If none does, you’ll learn to produce and be left alone with the judging.
How we work Workshops on your material
Our workshops start where the beginner courses stop. We work on your material, and you leave with two things: something that runs, and a yardstick for judging machine results before they leave the house.
Write to us with what your team is working on right now, and we’ll build the workshop around it. Watching first works too. On 17 August 2026, in the online course “Agents do the work”, we show on our own material what we measure machine results against before they become client work. Deliberately not a workshop but a 90-minute demonstration, in English, CHF 29.
The 139 events are a good sign, by the way. A country that fills this many AI courses is past the question “do we need this?”. The quality question will fill the programmes of 2027. Your team can answer it for itself this year.