200 Projects Later

In 2019 I started streaming on Twitch. I didn't have a plan. I had VS Code open, a vague idea for a chatbot, and maybe 5 people watching. Somehow that turned into 200+ open source projects built live on stream with a community of people I genuinely love. I still don't fully understand how it happened.

The pitch was simple: I build something from scratch every stream, and chat helps. Not "chat watches me code." Chat helps. People would suggest features, catch bugs, drop pull requests mid-stream, argue about variable names — the whole thing. And because we were starting fresh each time, nobody needed to have context from last week. You could show up for the first time and immediately be part of it. That mattered more than I realized at the time.

Some of those projects became real things. ComfyJS started as a way to make Twitch chat integration less painful, and it ended up being used by thousands of streamers. PixelPlush turned into an actual business. The whole "Comfy" ecosystem — ComfyDB, ComfyWeb, ComfyKit — grew out of stream ideas that people actually needed. We'd build something in two hours on a Tuesday night and then I'd get DMs a week later from someone using it in production. Wild.

And then there were the absurd ones. TrickOrTreatBot. ClippyAssistant — yes, we brought Clippy back and put it in Twitch chat, because of course we did. A stream notification system that played sound effects through a physical Raspberry Pi setup. A dating app for dogs. I'm not making that up. Not every project needed to be serious. Some of the best streams were the dumbest ideas, because the energy was pure chaos and everyone was laughing.

Here's the thing I took away from all of it: the viewer count never mattered. The best streams weren't the ones with the most people. They were the ones where 20 people were locked in, pasting code snippets into chat, debating architecture decisions, and losing their minds when the demo actually worked on the first try (rare). That taught me something real about building software. Ship fast. Don't overthink it. Let the people using the thing tell you what it should be. Have fun or what's the point.

I think about that era a lot. I'm not streaming as much these days — Microsoft keeps me busy, Prague keeps me happy, and I've got a dozen side projects pulling me in different directions. But those streams are the reason I think the way I do about software, community, and building things. Two hundred projects, hundreds of contributors, thousands of hours of live coding. No regrets. Not one.

If you were there for any of it — thank you. That community was the best thing I ever built, and I didn't build it alone.