TypeSafe AI, creator of Jev, has closed a $870 million funding round led by Andreessen Horowitz with participation from Sequoia and existing investor DCVC. The round values the company at $7.5 billion just weeks after Jev's market debut on September 15. According to TypeSafe's data, the model is already in use by roughly one-third of Fortune 500 companies.
Founded in 2024 by former OpenAI researcher Diogo Almeida, ex-Meta research engineer Sasha Sheng, and entrepreneur Erik Gafni, TypeSafe has built its business on a technical approach that diverges from conventional language generation. While Jev uses a transformer-based architecture, it does not operate as a traditional language model and does not produce text or code outputs. Instead, it calculates probabilities directly to emit what the company calls calibrated decisions, optimizing for process automation rather than dialogue generation.
The distinction matters for enterprise efficiency. As Almeida explains, AI systems excelled at human language over the past four years, but that capability doesn't translate well to automation because computers operate in a different language. Jev's design eliminates the token overhead typical of standard large language models, delivering computationally faster responses with lower resource consumption—a critical advantage in high-demand corporate environments.