Weekly Claude Build
I'm building a software project weekly using Claude Code. My North Star is the number of users I can get by the end of my final UChicago quarter (10 weeks long) - my goal is 1000.
The Initial Idea
I was talking to a friend about building today, and I gave him advice to just build something with Claude every week. That got me thinking - why can’t I do that?
The Motivation
Recently, I’ve been tinkering with ideas for how to figure out problems. I’ve also had an incredible time using Claude Cowork for my AI, innovation, and growth (a course at UChicago) project. In the AI age, I need to figure out how to best solve problems with AI in my pocket.
I can build things super quickly now, so what becomes more important? I hope to figure it out.
How I’ll measure success
I think measuring users (since I’ll be building software) will be a simple and strong approach. Each week I should ship something that someone else can sign on to and use. Revenue would be cool, but I think just creating something valuable for others and getting rid of the WTP filter will be a helpful place to begin. Remember - no one was going to pay for Facebook when it started.
How I’ll keep ideas flowing
My problem journal will become even more important when I’m building weekly. I find that with Claude Code, spinning up software is excessively easy. Clarity in building is harder.
While it’s tempting to “just start building,” I’m not a huge fan of this when it comes to working with Claude. Early design decisions can lock you into paths that aren’t optimal and eventually have to be backed out of (which Claude isn’t great at) to pursue a different path.
Taking time to build clarity, map the system, and then building is the proper path. I’ll probably hang out a lot in Miro boards to get this done.