RedAgent
A team of AI agents that attacks any AI chatbot, proves what breaks with Arize evidence, and verifies the fix — autonomous red-teaming with a closed loop.
Core trick
Runs attacker agents against a chatbot, logs failing traces to Arize, and re-tests the same attacks after a fix.