Guide Labs

Guide Labs develops interpretable and auditable AI systems and foundation models that humans can reliably debug, trust, and understand, addressing the fundamental black-box problem in modern AI. The company is building a new class of AI systems where model outputs can be explained and steered using human-understandable features, with capabilities including prompt attribution (understanding which part of the prompt is responsible for the output), token identification in pre-training and fine-tuning data, and concept-level interpretability. Guide Labs introduced Clarity, the first inherently interpretable AI platform, which provides transparency across prompt, concept, and training dimensions. The team brings over 20 years of experience in interpretable machine learning, with PhDs from MIT, UMD, and MILA, and has published 24+ research papers on interpretability at top ML conferences, including developing the first interpretable generative diffusion model and LLM. Guide Labs differentiates itself from mainstream AI approaches by building interpretability into the foundation model architecture itself rather than applying post-hoc explanation methods to black-box models, enabling developers to understand, debug, and steer AI behavior at a granular level. Founded in 2023 and headquartered in San Francisco, California, Guide Labs has raised approximately $18.8 million in a Seed round backed by Y Combinator, Initialized Capital, Tectonic Ventures, Lombardstreet Ventures, E14 Fund, and Pioneer Fund.

AI Infrastructure > Machine Intelligence Systems > Machine Learning Platforms > Deep Learning Platforms > Low Code Platforms > Generative AI

Company Wise > Google,

College Wise > Massachusetts Institute of Technology