Benchmark for measuring wasted AI-agent work — dead-end and confidently-wrong work normal observability misses.
By chatting or signing in you agree to the Terms and chat-message logging (revocable in History).