TypeSafe AI announced on its official blog on October 9, 2026, that it has raised an $870 million Series A at a $7.5 billion valuation, led by Andreessen Horowitz with participation from Sequoia Capital, existing investor DCVC, and a group of angel investors, with a16z partner Martin Casado joining the board. The round lands just 24 days after its decision model Jev launched on September 15.
Jev sells software a judgment it can use directly, not text
Jev is built on a transformer architecture but is not a large language model: it returns no paragraphs, only probabilities, which the company calls calibrated decisions. Software asks a question and gets back a yes or no, a pick from a list, or a score whose meaning developers can define, with no need to parse natural language down into structured fields afterwards. TypeSafe's pitch is that this runs far faster and uses far fewer tokens than an LLM, and that it is built for automating tasks rather than generating text or code. The company was founded in 2024 by Diogo Almeida, Sasha Sheng, and Erik Gafni. Almeida was previously a researcher at OpenAI who worked on InstructGPT-related research, and Sheng was a research engineer at Meta. On launch day the company also announced a $40 million seed round led by DCVC.
A third of the Fortune 500, read with care
According to the company's own blog, about a third of the Fortune 500 are already using Jev, and it has saved customers millions of dollars in production. The new money is earmarked for three things: more machine-native models, the rest of the infrastructure for building smart software, and the enterprise features customers have been requesting. This corner of the market is suddenly busy. We recently covered Microsoft turning judgment into a standalone model in Microsoft-Decision-1, and OpenAI earlier turned the same kind of judgment into an interface in its Decisions API public beta. TypeSafe's distinction is that it is an independent company doing only this, and that capital has now backed it at a $7.5 billion valuation less than a month after launch.
For buyers, the valuation and the adoption figure are both company claims. The two checks that matter are the familiar ones: whether Jev's probabilities hold up on your own workloads, and whether the latency and token savings beat the cost of adding and maintaining another vendor. This round turns not-writing-text, only-judging from a product shape into a heavily funded category. That the category now exists does not mean every customer should switch today.
