TypeSafe AI Emerges From Stealth With $40M to Put Decision Models Inside Software

The seed round was led by deep technology venture firm DCVC Management Co. LLC. Forbes reported that the round valued TypeSafe at $200 million, citing a person familiar with the transaction. The company was founded in 2024 by Chief Executive Diogo Almeida, who worked at OpenAI on reinforcement learning from human feedback, InstructGPT, ChatGPT and GPT-4. Co-founders Erik Gafni and Sasha Sheng also have extensive AI experience.

TypeSafe's first model, Jev, is built to return structured decisions instead of conversational text. It accepts structured questions and produces typed answers that other software can process, including yes-or-no probabilities, selections from a defined list and scores on a specified scale. Each decision carries probabilities and a confidence measure, which lets developers set thresholds for when an application proceeds automatically, gathers more information or sends a case to a person.

The company argues that the traits that make large language models good at talking to people work against them in production software. Such models can generate plausible but incorrect information, vary their methods between requests and state uncertain answers confidently, so applications that need consistent output often end up with human review in the loop. "We've been optimizing for humans, and we're superhuman at pleasing humans," Almeida told Forbes.

TypeSafe describes Jev as a "System One Model" trained with a method it calls Reinforcement Learning for Calibrated Decisions, an approach intended to make AI a composable software primitive that can be combined with deterministic code. According to the company, hundreds of decisions can be generated in parallel from a single prompt and assembled into larger workflows.

One example the company gives is insurance underwriting, where Jev could review available evidence about a property and estimate the likelihood that it will catch fire. A developer could let an automated workflow continue when confidence is high and route an ambiguous case to an underwriter.

TypeSafe says Jev uses parallel processing to deliver results in less than 100 milliseconds, making it up to 100 times faster and less expensive than other frontier models. Its website lists a cost of 39 cents per 1,000 workflows, against $3.31 for OpenAI's Gpt-5.6 Luna and R19.49 for Anthropic PBC's Claude Haiku 4.5. TypeSafe reported that its own tests found Jev nearly 194 times faster and about 445 times cheaper than the language models used for comparison. Those figures have not been independently verified and will vary by workload, network location and comparison method.

The constrained output does not guarantee correct decisions. TypeSafe says its use of probability ratings minimizes the risk of hallucinations, in which a model presents a wrong answer confidently. The company also says developers need to test how well Jev's confidence scores correspond to actual accuracy on their own data.

TypeSafe identified its largest opportunity in high-volume business processes such as classifying service requests, evaluating invoices, triaging security alerts and reviewing the results of AI agents. In those cases, a specialized decision model could act as a control layer while a language model handles tasks such as drafting text.

DCVC General Partner James Hardiman said TypeSafe is addressing "one of the biggest remaining challenges in AI" by making models reliable enough to embed in products at scale. Jev is currently available through an early-access waitlist.