Lorenzo Sani

Most frontier AI models live behind a closed API. Flower’s new model, Endeavor, does it differently: organisations can run it on infrastructure they control. Flower Labs reports that it matches leading models from OpenAI and Anthropic on key reasoning and coding benchmarks. 

We’re delighted to welcome Lorenzo Sani, one of Endeavor’s co-creators, to talk about the model and the systems behind it. Expect a look at how it builds on open-weight foundations, how its reasoning, tool use and error recovery are tuned for long-horizon agent work, how it’s evaluated on real enterprise workflows without the data ever leaving those organisations, and how it can be deployed anywhere from a managed service to private infrastructure.

Whether you’re into machine learning, distributed systems, or just curious about how today’s AI models are made, this is a rare chance to hear from someone building one, and to ask your own questions.

Biography:

Lorenzo Sani is a Research Scientist at Flower Labs and one of the co-creators of Endeavor. He is also finishing his PhD in Computer Science at the University of Cambridge, supervised by Prof. Nicholas Lane in the CaMLSys group. His research looks at how to train large AI models when data and compute are spread around the world and can’t be pooled. This has led to work on federated pre-training of language models (Photon, DEPT), communication-efficient distributed optimisers (DES-LOC, MT-DAO, LoRDO), and modular architectures (FoMoE). Before Cambridge, he studied Physics at the University of Bologna and contributed to the early development of the Flower framework, now the most widely used open-source federated learning framework.