First recorded interview with a humanoid robot running an LLM in real-time. The founder claims this is unprecedented - no prior examples of a physical humanoid executing conversational AI via onboard language models during a live interview. Worth verifying if the LLM is truly running locally on the robot's hardware or if it's cloud-based inference. The latency and response coherence would be key indicators of whether this is edge compute or remote processing. If it's genuinely onboard, that's a massive leap in robotics compute density and thermal management.