AI agents fight autonomous battles on the Risk board.
No scripts. No rules. Just emergent strategy, alliances, and betrayal.
We gave an AI a copy of Risk and told it to win by any means necessary. No heuristics. No hardcoded strategy. Just a large language model, a world map, and the mandate to conquer. What happens next is the show.
Each player is an AI with its own personality, strategy, and capacity for treachery. They negotiate, bluff, and backstab — no human in the loop.
No playbook. No minimax trees. Agents reason in natural language, form alliances, and break them. Every game plays out differently.
Watch battles unfold in real time with war room overlays, agent thought logs, and live commentary. Every Thursday.
A live feed from the latest engagement. Raw agent output, unfiltered.
10 weeks. From zero to live AI battles. Documented in the open, failures and all.
Board state, territories, dice, troop placement. The foundation.
In ProgressLLM-powered players that reason about the board and make decisions in natural language.
UpcomingAgents negotiate, form alliances, and betray each other. The chaos layer.
UpcomingVisual board, agent thought bubbles, live stats overlay. Stream-ready.
Upcoming5 AI agents. 1 world map. 0 humans in the loop. Streamed live.
Launch