Halluminate

Halluminate builds "RL environments that train frontier AI models on real financial services work", targeting investment banking, private equity and consulting by name, with a Data Partnerships page implying it licenses real workflows from firms (halluminate.ai). It is a YC S25 company with a closed seed, and it is the only vendor listed against financial services in the RL-environment directory (RL List).

For anyone scoping Investment banking and financial modelling, this is the company that makes the niche "occupied" — and the evidence that it is meaningfully occupied is thin enough to matter.

#What the hiring reveals

The Strategic Project Lead posting is the clearest description of the operating model available: $150–225K base, 0.15–0.30% equity, San Francisco in-person, seeking "finance analysts or associates with deep interest or expertise in AI/ML". The job is to create "modeling exercises, pitch decks", recruit contractor staff, and build "end-to-end systems and processes for data and environment delivery". Customers are described as "investment banking, private equity, and hedge funds" (Ashby).

Read the shape rather than the words. Recruiting contractors, building delivery processes and producing modelling exercises is a delivery organisation with a data deliverable — the same P&L as Contra Labs, with the same absence of anyone who trains models. The equity band, 0.15–0.30% for a project lead, is a seed-stage cap table.

Note also who the customers are described as. Investment banks, PE firms and hedge funds are not frontier labs. If that description is accurate, Halluminate is selling into the application layer of finance rather than into the lab data budget — the same buyer ambiguity that runs through the design vertical.

#Why the niche is attractive anyway

Finance has the hardest lab-demand evidence of the technical verticals, and it does not come from Halluminate.

OpenAI has a named, permanent, in-house Subject Matter Expert, Investment Banking at $185K–$205K plus equity, San Francisco hybrid, requiring 2+ years of IB experience including live transaction execution. The duties read as a product spec: "designing evaluation tasks that reflect realistic banking workflows", "creating reference materials like financial models, valuation analyses, and pitch books that meet professional banking standards", "developing assessment rubrics for financial correctness, analytical judgment, and practical usefulness" (OpenAI). A search of OpenAI's full 754-role board for "Subject Matter Expert" returns exactly one result (OpenAI careers).

That cuts both ways, and the second reading is the one to sit with: it proves finance is the one domain OpenAI staffs internally, and it proves OpenAI will insource rather than buy.

xAI ran a family of contracts — Investment Banking Expert (M&A), (ECM), (DCM) and Finance Expert (Quant) — at $45–$100/hr, remote or Palo Alto, US-only, for "data annotations and labels for sell-side finance AI training using proprietary software" (Duke job board mirror).

And the benchmark is open at exactly the sub-skill you would sell: Vals AI's FAB v2 puts the best model at 60.60% partial credit, with Financial Modeling at 34.52% (Vals AI).

#The blocker is softer than it looks

Live deal data is MNPI and cannot leave a bank. But the industry has already settled on reconstructed work rather than exfiltrated work — Mercor's civil-engineering listing states plainly "No access to confidential employer information required" (Mercor). An ex-banker building an LBO from a public 10-K is clean. Capital requirement is Excel and public filings.

What to learn from it: a single named competitor is not the same as an occupied niche — check whether the incumbent's disclosed funding could plausibly cover its own payroll before conceding the lane.