Guide Labs

Guide Labs is a San Francisco-based startup developing a new generation of inherently interpretable Large Language Models (LLMs). Founded by CEO Julius Adebayo and Chief Science Officer Aya Abdelsalam Ismail, the company's core innovation is an LLM architecture featuring a 'concept layer' that allows every generated token to be traced back to its training data origins. This approach addresses the 'black box' problem of traditional deep learning, enabling greater transparency, control, and reliability for AI applications across consumer, regulated industry (e.g., finance), and scientific research sectors. Guide Labs open-sourced its 8-billion parameter model, Steerling-8B, demonstrating 90% of frontier model capabilities with less training data.