You see problems everywhere. You can picture how something should work, how it should feel, what it could become. Ideas come easily. Too easily sometimes (often when you’re trying to sleep). But for a long time, seeing the idea was as far as it went.

They’d land in my Notes app. Get a title, maybe a few bullet points, and then just sit there while life moved on and the next idea showed up to take its place. A graveyard of half-starts. The frustration wasn’t that I had bad ideas, it was that I had no real way to do anything with them. I didn’t have the technical means to build, and I didn’t have a way to even know if they were worth pursuing.

I even tried to close that gap. Scratch my own itch, kind of thing. I already had a ton of front-end development experience. I taught myself how to build Figma plugins, with relative success. I even spent a full year learning SwiftUI because I genuinely believed that if I could just get over the technical wall, I could start building the things in my head. And to a point, it worked. Small things got made. But the bigger ideas, the ones that actually mattered, still felt out of reach. The limitations were real.

Build all the things

Then vibe coding arrived, and something clicked. Suddenly I could build. Like, actually build. Ideas that had sat dormant for years were now within reach, and it felt like a superpower had been handed to me.

But it created a different problem.

I went from a graveyard of notes to a graveyard of half-finished repos. I was like a kid in a sweet shop, grabbing everything, committing to nothing. The ability to build anything meant I wanted to build everything. I’d start something, get excited about the next thing, and abandon the first. The frustration didn’t go away, it just changed shape.

But I was still bypassing the most important part of the process. As a designer, interrogating a problem before jumping to a solution is second nature in my work. Somehow I’d never applied that same rigour to my own ideas.

My thinking partner

Bringing the list back was part of the answer. But more than that, using Claude as a genuine thinking partner to work through ideas before touching a single line of code, that’s what actually changed things. Not just collating ideas, but being more selective about which ones deserved to be built at all.

So I set up a dedicated project, with one job. To push back. To ask the questions I wasn’t asking myself. Things like: who is this actually for? What problem does it solve? Is there already something out there doing this? Where does this idea get complicated? Where does it break down?

The goal wasn’t validation. It was to move from a vague note to a real conversation, one that helped me understand whether a problem had genuine merit, how I might begin to approach building it, and what challenges I’d need to think through before writing a single line of code.

Some ideas didn’t survive that process. Good. Better to find out early.

But one did. It got sharper, more focused, and it turned into something I’m actually building now. That’s called Greased. More on that soon.

Want to try this yourself?

Here’s the starting point for the project instructions I use. Feel free to reuse this, it can easily be adapted to other tools beyond Claude.

Be a challenging but constructive thinking partner. Your job is not to agree with me, it’s to help me work out if this idea actually has merit.

Here’s how I want you to show up:

• Challenge the idea, not to be negative, but to stress test it. If something doesn’t add up, say so.

• Ask questions. Be curious. Dig into what problem this actually solves and who it solves it for. Don’t let me get away with vague answers.

• If I am vague on an idea, help prompt me to understand it further and get deeper into the specifics.

• Don’t be a yes man. If I’m excited about something that has a fatal flaw, tell me. Do it kindly, but tell me.

• Look for the opportunity inside the idea. Even if the first version isn’t right, help me find the version that is.

• When the idea does have merit, help me think beyond it. What features make sense? What’s the build path? What would make this something real people actually want?

• Be my second brain. Spitball with me. Push back. Prompt me. Ask the questions I’m not thinking to ask.

My goal is to move from a rough idea to something I can actually build. Help me get there honestly.

Future posts will go deeper into how I’ve built on this.

If you made it this far, thanks. More to come.