Halluminate has raised $30 million in Series A funding to expand its work developing reinforcement-learning environments and benchmarks intended to improve artificial intelligence performance in financial services and other forms of knowledge work. Oak HC/FT led the round, with participation from Y Combinator, Orange Collective, FT Partners and Heavybit.
The financing brings Halluminate’s total capital raised to $38.5 million.
Halluminate describes itself as a data research lab focused on building benchmarks and reinforcement-learning environments that can help frontier AI models perform complex professional work.
The company’s initial focus is financial services, with plans to expand into consulting, insurance, operations and adjacent knowledge-work categories.
Halluminate argues that while current AI systems have made rapid progress in areas with objectively verifiable answers, they still struggle with professional tasks requiring judgment, nuance and complex deliverables.
Examples cited by the company include building realistic financial models, creating client-ready presentations and producing well-balanced strategic analyses.
Halluminate said it has grown from zero revenue to a mid-eight-figure revenue run rate over the past 10 months while remaining strongly profitable.
The company has also built and scaled reinforcement-learning environments with four of the five leading closed-source U.S. AI labs.
Halluminate recently released Westworld Due Diligence, a benchmark designed to test AI systems on realistic financial work.
Its team includes former founders, researchers, physicists, domain experts and engineers with experience at Meta, Scale AI, Capital One Labs, McKinsey and Goldman Sachs.
The Series A will support Halluminate’s effort to continue building data, evaluation systems and training environments that can improve AI performance on complicated professional tasks.
