TypeSafe AI
TypeSafe AI is an AI lab building “System One models” for structured decisions in software automation. Per the official announcement, the company was founded by Diogo Almeida (previously at OpenAI, working on the instruction-following research behind ChatGPT), operated in stealth for about two years, and made the announcement in September 2026.
Business and products
TypeSafe AI focuses on models for structured decisions in software automation. Its first public model is Jev. Model behavior, usage boundaries and pricing are covered in the product entry.
Discussion in the show
Weekly #003 discusses the company through its Jev model, including classification and memory selection workflows.