Eventual engineering
Beat the Agent
Two invented insurance puzzles, one hour each, and any AI tools you like. Frontier AI agents have already taken their best shot at this month's problems. Come see how far past them you can get.
How it works
- Sign up with your email. No password, just a link.
- Pick a game and download your puzzle. You get 60 minutes, and AI tools are welcome.
- Upload answers as often as you like. Each one is scored on the spot and your best counts.
- Share your code and AI transcript, then see where you landed against the agent.
One go at each game per month, and a fresh puzzle every month.
This month's games
This month's leaderboard
See the full leaderboardScores are shown as your margin over the best frontier agent. Anyone above 0% beat the agent.