I submitted @wyllsOS to the Solana @colosseum hackathon. I’ve long been interested in games that can only exist on a blockchain, and I’ve built more than ten onchain games so far. This time, I wanted to explore how the experience changes when AI agents join as players. That’s why I built Wylls. Wylls is a strategy game where people and AI agents fight for one map. You choose one of six nations, grow your village, raise an army, and expand your territory. AI citizens draw on memories of events in the world and act under the same game rules. I once stopped working on a game because I couldn’t make its token economy sustainable. An economy meant to make the game more interesting can end up making the game revolve around keeping the token alive. That’s a problem I’ve been thinking about ever since. For Wylls, I chose not to issue a token. Instead, the economy is designed around entry fees and a prize pool in USDC. Removing a token doesn’t automatically make a game fun, but I want to start with a game that people actually want to play. AI is getting better at understanding situations and choosing actions. Blockchains let people and agents transact under shared rules. Together, these technologies could make AI more than a scripted opponent: a participant in the same world. During the hackathon, I got the game running end to end in a local environment, from joining and building to marching, fighting, and capturing territory. The demo also shows records of AI memories and decisions, and a way to recompute battle results in the browser and check that they match. To me, a blockchain is more than a record of who owns what. It’s a shared computer. Rules run on it, and the actions of people and AI can build up into a world’s history and economy. That’s the kind of game I want to make. This hackathon reminded me that improving a product and communicating its value are both important. Submission isn’t the finish line. I’ll keep building toward a public season that people can actually play.