Skip to content

Max out Claude Code every week until I make 10K a month

3 min read#ai#startup

// part of derkinzi.de

The other day, building Pepelly and maxing out my Claude tokens for the week, I had a thought: years ago, it took you a month to build and launch an app. As an MVP, basically a crude one. And I think these days, with marketing, you can easily do it in a week.

So the whole idea is: max out my Claude Code subscription every week until I make 10K a month.

I tweeted about that the other day. It didn't get any traction, but I still kind of want to do it.

Where I'm at

I'm currently in the process of preparing for this, and gathering ideas. I'm sure I can already wiggle a few of my ongoing projects in there. This was also the reason for relaunching this whole website, to have it a bit as a hub for this: every project gets its own page with a changelog, and whenever inspiration strikes, I'm gonna build a new one.

I'm not sure if I can keep the one a week up. Also because there's stuff like ContentIn that is actually working already and needs some attention from time to time, or some more attention from time to time. But I'll also share little tricks and stuff that I'm doing to actually get these things working.

Ground rules

  1. No AI generated ideas. Whatever I'm gonna build will be an original idea.
  2. A very, very big MVP, or rather what's already called a 1.0, instead of something crude to test the ideas. Because again, building doesn't cost anything anymore.
  3. Always launch a full project: a working product with marketing attached to it, and ideally an idea for ongoing marketing. For the last projects that was always SEO, and I'm basically building agents and strategies to just write and get content out, to get some initial traction.
  4. Apps, mobile apps, micro SaaSes, websites. I don't want to launch services or info products. Ideally something that works tech only.

What I want to build

  • Maybe rebuild, basically reverse engineer, an existing SaaS.
  • I would love to do something that's somewhat real life connected, hardware based, whatever, to add an additional moat. Don't really have a practical idea for it yet, though.
  • It would be cool to build a game, maybe for my daughter. I've never really built a game before, ever. So that would be very interesting. Probably that one could take longer than a week, just because I have no idea about that, and I have no idea how good AI is with it.

Two schools

The whole idea behind this: there's two schools out there.

One says AI is supercharging the age of the builder. I mean, I always was a builder, and with AI I can just do stuff that I couldn't have dreamed about, and I can do it so much easier than ever before. This whole thing would be a test on how good you can actually use AI. Not only to build stuff, that should be fairly simple, but also for marketing and so on.

The other school says SaaS will be obsolete in the future, because everybody can build their own stuff. I do agree to this to a certain extent, but I'm also sure that there's always need for some software. Or you try to build the shovels, which could be data MCPs, I think that's a very big shovel, or service MCPs that will never be part of the AIs themselves, more like connected tools.

So no, I don't think the age of the builder is done. But definitely the world is changing, and we need to figure out what works now, really.

I'll log every launch here. Week one is Pepelly.