Guide Labs

Guide Labs is a product focused research company building interpretable and auditable AI systems and foundation models that humans can debug, understand and steer. Its first product is Clarity, introduced in June 2026, which the company describes as an inherently interpretable AI platform. A featured demonstration of prompt concept training shows which part of a prompt is responsible for a given output. The company aims at teams and domain experts who need to trust model behavior in consequential decisions, where explanations from conventionally trained models can be unrelated to how the model actually decides. Its team reports more than 20 years of combined experience in interpretable machine learning, with doctorates from MIT, the University of Maryland and Mila, and more than two dozen papers at leading machine learning venues. It states that it developed one of the first interpretable generative diffusion models and an interpretable large language model, trained to explain outputs through human understandable factors, and has applied the approach to protein property prediction. Access to Clarity is requested through the site. No pricing is published, and no pricing page was found, so terms need to be confirmed with the company. The listing sits in a coding category, but the product is a general interpretable AI platform rather than a coding assistant.