What a Packed AI Coding Event Says About AI Adoption
2. Juli 2026
One developer. 20 days. A working loan-check application for a bank.
That is what caught my attention at last night's Claude Code community event in Zürich, hosted by Adnovum. And it was just the opening story.
Seats were limited. You had to apply. The room was packed anyway. UBS, Google Cloud, Meta, Deloitte, AXA, Zurich Insurance, ETH. And so many women! That alone made my evening. It was lovely catching this moment Sofie Maltha Biehle 💛
Three talks, three things I am keeping:
→ Adnovum showed four real projects done with Claude Code, from greenfield to legacy modernization. In all the projects, the work moves upstream. The human work moves to before the coding. People define what to build, how to check it, and what the AI must not do. The AI writes the code. Implementation starts later but finishes faster. One speaker said it best: our job now is to get comfortable spending more time not watching it produce code.
→ BugBounty Switzerland on security: AI made finding vulnerabilities cheap, so the flood of reports will not stop. But in the flood, you need to understand context. As they say, security cannot be proven, but you can prove you went looking hard in the right place before someone else did. (Also, isn't BugBounty the coolest name for a startup on security!)
→ Google Cloud on regulation: FINMA, DORA and GDPR get whispered like ghost stories. Read the actual rules and most of the fear dissolves. Understand regulation before panicking.
The talks ended and nobody left. People stood around with drinks, asking the speakers questions, trading business cards. That told me more about where AI adoption stands than any survey I have read this year.
Have you been to any cool AI event yet this year?
The human work moves upstream
The pattern in every project shown that evening was the same. The human work has not disappeared. It has moved earlier. People define what to build, how to check it, and what the AI must not do. The machine writes the code. Implementation starts later and finishes faster.
That is a leadership story more than a coding story. Defining what good looks like, setting boundaries before the work starts, deciding what a system must never do: this is judgment work, and it is exactly the part AI hands back to humans. The teams that struggle with AI are usually not short on tools. They are short on clarity about what they actually want from them.
The other signal was the energy in the room. Nobody left when the talks ended. People stayed, asked questions, traded cards. Adoption driven by curiosity looks very different from adoption driven by mandate, and it moves faster. I see the same split inside organisations: where people pull AI in, it spreads. Where it is pushed down from above, it stalls.
That move from AI pressure to AI confidence is a path I walk with leaders one by one. I described what it looks like in practice in this Private AI Advisory case study.
If you want to bring this conversation into your own team, I have a free What AI Can't Hear pack with the seven listening behaviours from my TEDx talk plus ten practical questions you can use to talk about AI without fear or confusion. Leave your email here and I will send it over: https://www.thechangerepublic.com/free-resources