The signature keynote on AI in the age of intelligent systems — tailored to every audience.
The case I make on stage is that most of the value AI creates is not captured by the people who use it like a faster search engine, and not by the people who refuse to use it at all. It is captured by the people who learn to be its complement — who do the work that surrounds the model: framing the question, providing the context, judging the output, integrating the result into something real.
I borrow an analogy from any working kitchen. A senior chef does not slice the tomatoes. They don't decide what heat to set on the hot plate. They design the menu, source the ingredients, hire and train the team, set the standard, and taste what comes out before it reaches the table. The cooking still happens — it just isn't theirs to do. AI is doing the same thing to knowledge work right now, and the people who reorganise their day around the human work — framing, judging, integrating — are the people compounding fastest.
Every keynote is tailored to the audience in front of it. The spine of the argument is the same; the examples, the stories, and the lines of inquiry change to match the room. Forty-five minutes typical, longer with Q&A. German, English, or French.
Depending on what the audience needs, the keynote goes deeper on one or more of the following lines. Each one is a real working area that I keep current through my own practice; none of them is a prompt-engineering primer, because prompt engineering is the easy, boring part. The interesting work is upstream and downstream of the prompt.
Not how to prompt — how to think about the interplay. Where in your operating model does AI compound, and where does it produce expensive noise? Which decisions does it accelerate and which does it quietly degrade? The strategic question is rarely "should we adopt AI." It is which work to keep human, which work to delegate, and how to design the joint system so that the human side actually gets stronger over time rather than weaker.
Used well, AI is one of the best reflective tools a serious professional has ever had access to. It will tell you what your thinking looks like from the outside. It will surface assumptions you have stopped noticing. It will let you rehearse difficult conversations, stress-test your reasoning, and identify the gaps in your own arguments before they meet the world. Most people never use it this way. The ones who do compound at a different rate.
There is a conversation in founder circles about delegating less to people and more to AI — and, as a result, staying much closer to the actual details of the work. The image is the founder who still walks the factory floor years after they have achieved generational wealth. Same instinct. AI lets a senior professional drop two levels of abstraction back into the work without paying the time cost that used to make that impossible. What changes when you can.
When the conditions change as fast as they are changing now, the right move is not to do the old thing harder. It is to redesign how the work is done from first principles, using whatever singular advantages you happen to bring to it. The Fosbury Flop was not a better high jump. It was a different jump. Most of the productivity advantage available right now is sitting in differences like that, available to anyone who is paying enough attention to notice them. The talk is partly an argument for that kind of noticing — and the deliberate practice that makes it possible.
Which of these lines lands hardest depends on who is in the room. I'll ask before I build the talk.
Closed-door work for smaller groups, with real preparation in advance, organised around a concrete topic the team is working through — not a longer keynote with more time for Q&A. The shape is more like a working session with a facilitator who has built the kind of system you are considering.
Most often: boards, leadership teams, and divisional summits. Two to four hours. I work with the organiser to scope the brief beforehand, review the relevant material, and arrive prepared on the specifics. People leave with shared vocabulary, a clearer view of where their current AI work is overinvested and underinvested, and a specific list of next moves.
Sometimes the signature isn't what your audience needs. In that case I listen to the brief and build a talk to it. Past examples include AI in transformation programmes, hiring and team design in an AI-augmented company, and personal productivity for senior individual contributors. If your brief doesn't match the standard signature, send it anyway. The spine of the argument is durable; what changes is what gets put on it.
I travel for European engagements regularly and for selected international ones. Keynotes and bespoke talks are delivered in German, English, or French. Virtual delivery is available; in-person is preferred for executive sessions.
Or stay close to the work. I post the kernel of every essay on LinkedIn before it arrives here.
Follow on LinkedIn →