TypeSafe AI, a startup founded in 2024 by former OpenAI researcher Diogo Almeida along with ex-Meta engineer Sasha Sheng and entrepreneur Erik Gafni, has quickly made waves with its recently launched AI model called Jev. Unlike popular large language models that generate text, Jev operates by producing “calibrated decisions,” or probabilities, optimizing it for automation tasks rather than text or code generation. This approach reflects the company’s belief that traditional language models focus too much on human language, which is less effective for computational automation.

Since its September 15 debut, Jev has experienced rapid adoption, with TypeSafe claiming that a third of Fortune 500 companies have integrated the model into their operations. The AI community and large enterprises alike are impressed by Jev’s efficiency, noting it processes information significantly faster than conventional large language models while also utilizing fewer tokens. This token economy translates into reduced computational overhead and faster decision-making, addressing major bottlenecks in AI deployment at scale.

Backing this momentum, TypeSafe AI recently secured $870 million in funding at a valuation of $7.5 billion. The financing round was led by renowned venture capital firm Andreessen Horowitz and included participation from Sequoia Capital and DCVC, a returning investor. This sizeable investment reflects intense investor confidence in TypeSafe’s novel approach to AI and its potential impact across multiple industries seeking automation solutions.

TypeSafe positions Jev as a unique evolution in AI technology, moving away from generating natural language toward enabling smarter, faster automated decisions. With industry-leading engineers at the helm and strong venture backing, the company is poised to challenge the dominance of text-centric AI models and further accelerate enterprise adoption. The rapid rise of Jev underscores growing demand for specialized AI designed explicitly for operational automation rather than conversational tasks.