A San Francisco AI startup that came out of stealth in mid-September is now worth $7.5 billion. TypeSafe AI, maker of a model called Jev, has raised $870 million in a round led by Andreessen Horowitz, with Sequoia and existing investor DCVC joining, TechCrunch reported Friday.
The jump is steep. When TypeSafe announced itself on Sept. 15, it had a $40 million seed round led by DCVC, and Forbes put that valuation at $200 million, according to SiliconANGLE.
What Jev does differently
Most of today's popular AI tools, like ChatGPT, are large language models that answer in text. Jev is built on the same transformer technology, but it doesn't write sentences. Software sends it a structured question and gets back a typed answer, such as a yes-or-no probability, a pick from a fixed list or a score on a scale, along with a confidence measure. The company calls these "calibrated decisions."
The pitch is that programs can act on those answers directly. SiliconANGLE gave the example of insurance underwriting: Jev could estimate how likely a property is to catch fire, let a workflow continue automatically when it's confident and send murky cases to a human underwriter. TypeSafe says the biggest uses are high-volume chores like sorting service requests, checking invoices and triaging security alerts.
TypeSafe also claims Jev is far faster and cheaper than chatbot-style models. In its investment announcement, Andreessen Horowitz described it as roughly 1/100 to 1/500 the cost of frontier models and 100 times faster on classification tasks. Those are company and investor figures; SiliconANGLE noted the startup's own speed and cost comparisons had not been independently verified.
Fast adoption, by the company's count
The startup told TechCrunch that a third of Fortune 500 companies already use Jev. Andreessen Horowitz put the figure at 25% and said the model generated 1 trillion tokens in its first three days.
TypeSafe was founded in 2024 by CEO Diogo Almeida, who worked on InstructGPT, ChatGPT and GPT-4 at OpenAI, along with former Meta research engineer Sasha Sheng and engineer Erik Gafni. "We have been super good at human language for four years, but it's not useful for automation because computers speak a different language," Almeida told TechCrunch.
Why it matters here
The round adds TypeSafe to the list of San Francisco AI companies drawing huge checks this year, and it's a bet on a different direction from the big labs: smaller, cheaper models built into ordinary software rather than chatbots people talk to. Whether its adoption claims hold up will show in how widely Jev spreads beyond early access.