TypeSafe AI is a San Francisco artificial intelligence company that builds System One models: models that return typed, calibrated decisions such as a choice, a score, or a probability for software to act on, rather than generated text. Its first model, 📝Jev, is sold to developers through a usage-metered API with Python and JavaScript SDKs.
TypeSafe was co-founded by chief executive Diogo Almeida, a former 📝OpenAI researcher who worked on InstructGPT and the reinforcement learning from human feedback behind ChatGPT, together with Erik Gafni and Sasha Sheng. After roughly two years in stealth, it launched Jev on September 15, 2026 alongside a $40 million seed round led by DCVC, at a reported valuation of about $200 million. The company argues that models trained to please human raters are poorly suited to the high-volume, repeated decisions inside software, and trains its models for calibrated decisions instead. Access began through an early-access waitlist, and later releases added image input and object localization (📝Jev Images). Reports in October 2026 of a much larger follow-on round, including a Bloomberg-attributed figure of about $870 million at a $7.5 billion valuation, have not been confirmed by the company.
We use Jev as a classify provider inside 📝MythOS Loops, where a fast, cheap yes, no, or pick-one decision beats asking a full language model.