Exclusive: Nine-Person Halluminate Raises $30 Million, Counts Four Top U.S. AI Labs as Customers

Halluminate, a nine-person startup based in San Francisco, recently raised $30 million in a Series A funding round led by Oak HC/FT. The total funding raised by the company is now at $38.5 million. The startup specializes in building AI training environments for financial work, with the goal of identifying areas where AI models fall short and creating simulated training environments to address those gaps.

Founded in 2024, Halluminate believes that AI training data and environments will become increasingly specialized by industry. CEO Jerry Wu describes the systems the company creates as "verticalized data research labs," highlighting the company's focus on verticalization. A benchmark released by Halluminate in August showed that seven frontier models achieved an average score of 51% on a simulated company-acquisition due-diligence process, highlighting the challenges in complex financial workflows.

This benchmark, along with the growing demand for post-training infrastructure in the industry, has attracted the attention of four of the top five closed-source U.S. AI labs as customers. Halluminate's AI training environments are expected to become increasingly complex, with the complexity doubling roughly every six to eight months.

This "Moore's law of environments" is crucial for pushing the capabilities of frontier models. The company's investors include Y Combinator, Orange Collective, Heavybit, and individual researchers from Anthropic, OpenAI, and Meta.