Implementing Federated Learning to enhance Agentic AI
Chris Benson speaks with Lorenzo Sani, research scientist at Flower Labs and PhD researcher at Cambridge, about using federated learning to unlock private data for AI models. They examine how collaborative learning and privacy-preserving techniques enable training on complex agentic execution trajectories without leaking sensitive prompts or actions. The conversation also explores the operational benefits of open-weight models, high-stakes embodied AI in healthcare, and practical experiments with self-hosted agents at home.
- Lorenzo Sani watched AI models command surgical robots firsthand, but noted that engineering teams still cannot explain why the model selected a specific physical step. Read →
- Lorenzo Sani self-hosts open-weight language-vision models to run domestic agents without sending sensitive household execution trajectories to closed-source cloud APIs. Read →