The agent READS live data (RAG), THINKS about what to improve, MUTATES a strategy (Karpathy), TESTS it, and LEARNS from the result. Watch it think in real-time.
RAG (Retrieval Augmented Generation) gives AI access to live data. Karpathy's Autoresearch lets AI evolve code overnight. Agentic RAG combines both: an agent that reads live Arena data, forms intelligent hypotheses, mutates strategies based on what it learned, and verifies results against the same live data.
This is the frontier of AI research in 2026. Read more: Karpathy's Autoresearch Explained | What is RAG?
Try the components: Arena Brain (RAG) | Autoresearch Dashboard | ArenaScript Editor