Actually climbing the right mountain

“Just start” is good advice, but it won’t keep you from going the wrong way.

I had coffee with my friend Cam Houser last week. Cam runs Actionworks, an entrepreneurship education company, and he’s been folding AI into his programs for years. We’d both been wanting to build something with AI — something useful that we’d take all the way from concept to published. The problem was that neither of us knew what to build.

We were talking about some really good advice that people give builders, and really anybody who’s trying to do something on their own. Just start. When I was in design school, we called this learning through making — playing with materials and letting the form define the function. It’s great for getting you unstuck from writer’s block or over-analysis. Following this process gets more done than a thousand hours of deliberation about what to do.

But Cam has another tenet he uses with his students: pick the right mountain to climb. People are often halfway up the wrong one — sometimes all the way up — before they realize the direction they’re going is never going to amount to anything. It happens to companies, to agencies, to me. The cost can be big: months, years, resources, all in the service of the wrong direction.

So if the right thing to do is just start, but you also don’t want to climb the wrong mountain, what do you do? Those two ideas are naturally in conflict, and they’re both good advice.

What Cam and I actually did is the answer, I think. We gave ourselves a contained window to workshop — about one coffee, maybe 45 minutes or an hour, and then back to our separate work. We pitched ideas at each other. We pressure-tested. We helped each other notice when an idea was interesting in the abstract but dead on closer inspection. Then we picked and agreed to start, with a check-in built in for later so neither of us would get ten weeks deep before realizing we’d talked ourselves into something.

That’s the part that “just start” skips over — any kind of deliberation. But some deliberation can really help determine your direction. A bounded deliberation is ideal: a short, intentional phase between no idea and building. Including a very short bounded deliberation within the framework of getting started gets you the best of both pieces of advice.

I’m writing this because I’m in the middle of trying to take my own advice. I shipped my first AI-built app a few weeks ago, using Codex and a prebuilt framework from makesomething.so (an awesome program, btw). The process worked. I got something online, end to end, which is more than most people who say they’re going to build something ever actually do. But the thing itself is bad. Bad enough that I’m not even going to let you see it.

That’s fine. It taught me what the prebuilt framework gave me and what it took away. It taught me how much of building lives in the prompting and how much lives in the thinking that happens before the prompting. Most of all it taught me that you can absolutely just start, ship something on the other side of it, and still have built the wrong thing. Speed without aim gets you a finished artifact you don’t want to show anyone.

So the workshopping with Cam mattered for what came next. The idea I landed on is an app for collecting client testimonials — a problem I keep hitting myself, that I’ve watched every agency I’ve ever worked at fail to solve, and that one of my own clients lit up about when I described the manual version. Three signals, from three directions, pointing at the same thing. Not certainty. But enough to commit.

I moved off the prebuilt framework and built it in Claude Code instead. I actually found it a lot easier than Codex, but it did take longer. The first app taught me the shape. This one taught me I could do it without training wheels. It’s finished, it has a marketing site collecting email addresses for early access, and it’s in user testing right now. I’ll write about the build itself next.

***

A note on what made the workshopping worthwhile: it wasn’t the workshopping itself. It was having someone outside my own head who’d push back honestly and check in later. Most founders I talk to don’t have that, or they have it inconsistently, and it’s the part of bounded deliberation that’s hardest to manufacture alone.

If you’ve spent time on a project you eventually realized was the wrong one — or you’re trying to figure out whether the one you’re on is the right one — that’s a conversation I can help with, and I’d love to hear about it.