Handshake Reinvents as Expert Data Provider
Handshake vs Mercor
Handshake’s shift matters because it turned a slowing campus recruiting network into a supply engine for scarce AI training labor. The company already had verified access to students, graduates, postdocs, and PhDs, so when labs started paying for math, law, and science experts to score model outputs, Handshake could monetize the same identity and credential graph in a new workflow, as hourly expert data work rather than job matching alone.
-
The concrete change was from selling software to career centers and recruiting access to employers, into selling expert hours to AI labs. By August 2025, Handshake was estimated at $280M annualized revenue, with about $80M from Handshake AI and $200M from the legacy recruiting business.
-
Mercor built this market from the other direction. It recruits and vets experts specifically for AI work, using interviews and tests to match doctors, lawyers, and PhDs to labs. Handshake started with an existing education centered network, which lowers expert sourcing friction because the credentialed supply was already aggregated.
-
This is part of a broader pattern. Invisible moved from virtual assistant work into RLHF, and Prolific evolved from academic research participants into AI evaluation. The common play is taking an existing human supply network, then repackaging it for model training, evals, and safety work as labs need more specialized judgment.
The next step is moving beyond pure labor supply into infrastructure and repeatability. As expert data work matures, the winners will combine access to scarce humans with better workflow software, quality controls, and audit trails, which lets Handshake sell not just expert hours, but a more durable AI data product.