← Writing

Shipping is marketing

I launched a framework for AI agents on Solana last week. Fifteen production tools, real chain integration, encrypted key storage, async all the way down. I also launched a token for it the same day, on pump.fun, using the same wallet I’d used for the last experiment, which got me sniped inside a minute. So here’s the honest version up front: the tech was ready and I fumbled the launch like an amateur. Both things are true. They usually are.

The pitch is simple and a little stupid. Give a language model a wallet and a set of tools and let it operate on a chain that can actually lose your money. Everyone spent this year saying agents were the future. Almost nobody wired one to somewhere with real stakes, because a demo that runs once in a notebook is safe, and a thing that holds keys and signs transactions is not. I wanted the unsafe version. The framework is the unsafe version with seatbelts: a tool registry, a keyring with encryption, risk checks, the boring scaffolding that turns “the model did a trade” from a party trick into something you’d leave running.

It’s open source. You can pull it, uv sync, drop in a key, and have an agent that checks a wallet, finds tokens, and trades on Jupiter or pump.fun in about thirty minutes. None of that is the interesting part. The interesting part is the token, and I want to be straight about it because most people aren’t.

the token is the funding and the leash

pump.fun pays creator rewards now. That changes the math on building in public in a way nobody’s quite said out loud. For the first time the thing where you ship a commit, post the diff, and get feedback also pays rent. I’m not taking side work right now. I sit here sixteen hours a day writing code, and the fees from a token cover it. That’s not a flex, it’s just the mechanism, and the mechanism is genuinely new.

It’s also a leash. The second there’s a token, half your audience isn’t there for the framework, they’re there for the chart. They’ll call the tech next-level on a green candle and call you a farmer on a red one, sometimes in the same thread, sometimes the same person. The code didn’t change between those two tweets. The price did. People have the attention span of a fish and they will tell you the fish is your fault.

I’m not going to pretend that doesn’t get to me, because pretending is how you end up posting something deranged at 2am. The discipline I’ve landed on is dumber and more reliable than managing sentiment: ship the commit, post the commit, go back to the code. The commit log is the one marketing channel that compounds. Everything else, the videos, the stream, the threads, is decoration on top of the fact that the thing is actually getting built, in public, where you can watch the diffs land.

shipping is marketing

That’s the whole thesis and it fits on a sticker, which is why I keep posting it. In a feed engineered to reward the loudest take, the most contrarian move available to a builder is to just keep shipping and let the artifact do the arguing. You can’t fake a commit history. You can absolutely fake a thread. The market mostly can’t tell the difference in the moment, but the difference is the only thing that survives the moment.

People ask what it’s for, like there’s a clean answer. Right now you can build a trading bot, a research agent, a sniping tool, a portfolio tracker: the framework is lego, the tools are the bricks. Someone already used it to build a KOL tracker that runs on top without touching the core, which is exactly the point: the framework should disappear under the thing you built with it. I don’t know which of these becomes the thing everyone talks about. Could be me, could be a dev I’ve never met. I just know it’ll be built on something that already exists instead of from scratch, because nobody starts from scratch when the scaffolding is sitting right there, open, tested, and free.

So that’s where it stands. A framework that works, a token that funds it and occasionally tries to drive it off a cliff, and a guy posting his commits into a casino and calling it a strategy. It mostly is one. I’ll keep shipping and we’ll see what holds.