Before you build, drill (I meant prototype)

Last two posts I talked about understanding what customers need to fire and hire — and how to get the real story out of them. But knowing why to build something is one thing. Figuring out what to build is another quest entirely.

Most teams follow an agile process that's really just waterfall happening faster. PRD → Figma → tickets → code → test. People say we're in a new era with AI now. Honestly? This was already true 10 years ago. I know, because that's why I switched from frontend engineering to product. Writing code wasn't the hard part. Making the right decisions was — and we kept getting that late or wrong.

A side hobby of mine — I followed how highways are built in my country. Before a highway gets built, highway builders do something called geological exploration. They drill holes along the entire route — not to build the road, but to understand what's underneath. If there's a tunnel planned, they'll actually drill a small pilot tunnel first. Not to "build" the tunnel. To learn what the mountain is made of, so they can design the real one to last.

That's what prototyping should be.

Not a rough draft of the product. Not a "quick and dirty" version you'll clean up later. A deliberate investigation into the areas where you have the most unknowns.

Prototyping is learning. Prototyping is de-risking. Prototyping is understanding complexity so you can make decisions about it.

In product work, your "geology" could be anything. Maybe you're an early-stage startup and you don't know if anyone actually wants your idea. Maybe you want to use a library you've never used before. Maybe you have a huge customer base and you're about to introduce a controversial feature with no idea how they'll respond.

Each unknown needs a different kind of drill. But the principle is always the same: go deep where the risk is highest, before you pour concrete.

Yes, AI gives you a faster, more powerful drill. And honestly, it's incredible — working with AI makes me so much better as a PM and it's also way more fun. But a faster drill doesn't mean you don't need to understand the unknowns and the hidden complexity. You do.

Don't jump from idea to building. Drill and learn geology first. I meant prototype.

Thanks to Ryan Singer for inspiration on this one.

This is post 3/8 from my talk at Dev/Hack/Day, @BTC Prague 2025.