The best conversations at conferences rarely happen on stage. They happen during coffee breaks, at dinner, on the walk to the taxi. That's where people talk openly about what actually worked and what didn't — no slides, no marketing filter. And those conversations are almost never recorded.

That has bugged me for years. On stage you get the polished version; in the in-betweens you get the honest one. Our industry could use more of the second kind.

The data and AI world is running hot right now. Everyone's talking about LLMs, vector databases, agents, MCP. But in trainings and coaching sessions I keep seeing the same thing: most teams are still wrestling with the same fundamental questions they were five years ago. Which data model fits? How do we historize properly? Who gets to see what? The buzzwords change; the problems don't.

That's why I'm launching [antifragile data] — data, ai and other stories in September 2026. A podcast that picks up where the coffee breaks leave off. I talk with people from the field — data modelers, engineers, architects, decision-makers. And deliberately also with people from adjacent disciplines whose perspective we're missing. Each episode is 30 minutes. Long enough for substance, short enough for your commute.

Three formats are planned:

• Deep Dives — me solo on a topic, tight and opinionated.
• Guests & Stories — conversations with practitioners and thinkers, in German or English.
• Quarterly Format — every three months with Kim de Vries, a look back and forward at what's moving the community.

What it won't be: no product demos, no silver-bullet episodes, no guests pitching their book. There's enough of that elsewhere. This is about real experience, dead ends included. Once it turns into charts and dashboards, we're out.

The name, by the way, isn't random. Antifragile is Nassim Taleb's term for things that don't just survive stress but get stronger from it. That's what the data world increasingly needs. Systems that don't just endure change but learn from it. Teams that get smarter through failure. Data models that evolve instead of break.

If that sounds interesting: at podcast.tedamoh.com you can sign up for the newsletter. You'll hear when the first episode goes live. And if you have a story worth telling, there's a form to apply as a guest too.

See you in September.

— Dirk