Jev, an AI model that does not generate text, has helped its developer TypeSafe AI raise $870 million at a $7.5 billion valuation. The funding comes just weeks after the model’s release and amid the company’s claim that a third of Fortune 500 companies are already using it.
A fast rise after launch
Jev was released on Sept. 15 and gained popularity quickly, according to the report. TypeSafe AI’s fundraising round was led by Andreessen Horowitz, with participation from Sequoia and existing investor DCVC.
The reported enterprise uptake is notable because it arrived so soon after launch. TypeSafe says large companies are already putting the model to use, though the source does not provide further details about those deployments or the tasks they handle.
Built to produce decisions, not prose
Jev uses a transformer architecture, but TypeSafe says it is not a large language model. Rather than outputting words, it produces probabilities, which the company calls “calibrated decisions.”
That distinction shapes the way TypeSafe presents the product. The company says Jev is designed for automation, rather than writing text or code. Its case rests on the idea that a system producing decisions can be useful for tasks where generating a human-readable response is not the goal.
TypeSafe also claims Jev works significantly faster and uses far fewer tokens than large language models. These are company claims in the report; no benchmark figures or independent comparisons are provided there.
The company’s argument for automation
Co-founder Diogo Almeida described the gap TypeSafe is trying to address in comments to TechCrunch last month: “We have been super good at human language for four years, but it’s not useful for automation because computers speak a different language,”
His comment frames Jev as an effort to move beyond AI systems built around language. If computers need outputs that can support automated action, TypeSafe’s approach suggests that probabilities may be more useful than a passage of generated text for some tasks.
The source does not detail how Jev converts those outputs into particular actions, or which kinds of automation are already in production. The company’s positioning and reported adoption point to interest in that model, while leaving the specific use cases open.
Founders and funding
TypeSafe AI was co-founded in 2024 by Almeida, former Meta research engineer Sasha Sheng, and engineer and entrepreneur Erik Gafni. Almeida was previously a researcher at OpenAI.
The new $870 million round gives the company a $7.5 billion valuation, with support from Andreessen Horowitz, Sequoia, and existing investor DCVC. Combined with the claimed early use by Fortune 500 companies, the financing signals strong interest in TypeSafe’s alternative to text-focused AI.
Whether Jev’s decision-oriented output can support automation across a broad range of tasks remains to be seen. For now, the company’s central pitch is clear: build an AI model for situations where systems need to make or support decisions, rather than compose language.