Halluminate raises $30 million Series A led by Oak HC/FT
The company builds simulated professional workplaces where AI models can train on long, scored financial tasks.
Y Combinator, Orange Collective, FT Partners and Heavybit joined the round, which brings Halluminate’s cumulative funding to $38.5 million. Oak HC/FT’s thesis is that training data must include the work professionals perform across tools and documents, beyond short answers drawn from internet text.
Halluminate recreates settings such as a financial analyst’s email inbox, transaction data room and required deliverables. Models must gather information, reason through incomplete inputs and produce a result that can be scored, giving AI labs feedback on tasks with longer horizons than a single prompt.
Its Westworld due-diligence benchmark tests this approach on private-market transaction work. The company’s broader training environments are intended to teach economic reasoning through reinforcement learning, where completing the work produces the signal used to improve a model.
What the funding supports
- Develop vertical data research labs
- Deepen professional task complexity in training environments
Halluminate
Benchmarks and simulated workplace environments that help AI labs train and evaluate agents on financial analysis and other professional knowledge work.
Company profile and funding history