Berlin, Paris, Milano – and a question for you

Hey friend,

The past few weeks have been a blur — Berlin, Milan, then suddenly Paris.
I’ve been deep in the Antler accelerator, iterating, prototyping, and rewriting everything ten times over. Fast pace, long days, endless caffeine.

And now… I need your help.

If you’re a solo founder, agency owner, or SMB builder, I’d love your input for a short survey (takes 5 minutes). It’ll help shape what I’m working on next.

👉 survey link

After two weeks of nonstop iteration in Berlin, I flew back home to Milan to catch my breath. But before I even unpacked, I got a message from the Entrepreneur First team — “Hey, we’re hosting a small retreat in Paris tomorrow, want to come?”

So… I slept three hours, grabbed a flight at 6 a.m., and spent two days at the Builders Retreat with Entrepreneur First x Anthropic — just 30 founders from across Europe, hacking, building, sharing ideas.

The event was sponsored by Anthropic (Claude), so there were tons of credits, demos, and late-night coding sessions with people from all over the world.

It reminded me why I love this space — that mix of raw creativity, chaos, and shared obsession with making something real.

And if you’ve been following the journey — yes, I was accepted to Entrepreneur First in London, but decided not to join the full cohort this round.

The tempo wasn’t right, and I felt I’d learn more by building on my own for a while.

Still, I’m keeping close with the EF and Antler teams

Now, something I’ve been reflecting on lately — especially after talking to other founders: How do you actually build something that lasts in AI?

Something more than another wrapper or API bridge?

After weeks inside both accelerators and talking to investors, I keep coming back to two things.

First — the moat is the data.

Model companies will always have bigger compute, larger distribution, more power. You can’t compete there.

But high-quality, hard-to-access data — the kind that’s sensitive, fragmented, or built through trust — that’s where early-stage founders can build something defensible.

The real advantage isn’t compute. It’s permission.

And that moves sideways, not upward.

Second — it’s about connecting dots.

You can build something hyper-specific, like an AI lawyer or AI nutrition coach. But the real potential lies in systems that understand context: how decisions in one area affect another.

Startups that can connect those dots — between business, psychology, creativity, data — those are the ones that shape industries, not just tools.

That’s what I’m exploring right now.

So yeah — between Berlin’s chaos, Paris’ creativity, and Milan’s espresso breaks, it’s been a lot. But it feels like the right kind of movement.

If you have 5 minutes, please fill out the survey — it’ll help more than you think.

👉 survey link

Also, my private community just got a new guide on prompt and agent engineering — how to actually structure and run AI agents for real workflows.

If you’d like early access or want to join future calls, you can join the community here.

stay caffeinated.
lead the machine.
launch anyway.

— Miron