Mercor shifts to enterprise agent platform

This pushes Mercor out of the staffing bucket and into the enterprise software plus services budget. Instead of filling open roles or supplying raters to AI labs, Mercor is trying to map how a company actually works, turn that into agent specs, connect those agents to tools like ERP, SharePoint, and Slack, then score failures against company rules so each workflow gets better over time. That is much closer to building a company specific operating layer than running a marketplace.1 2 3 4 5

• Mercor’s enterprise product starts with workflow discovery, not recruiting. Its own materials describe AI interviews, enterprise data, and workflow mapping to find agent opportunities, then moving into deployment, evaluation, and optimization. That makes the first sale look like consulting plus systems integration, with talent supply as an input, not the product itself.1 2 4
• This is distinct from Mercor’s original AI lab business, where it matched vetted doctors, lawyers, and PhDs into RLHF work for labs and charged access fees plus recruiting fees. In the enterprise version, the scarce asset is not just expert labor, it is the company’s internal process knowledge, approvals, source systems, and quality bar.4 5
• The closest comparable is Snorkel’s move from data tooling into building specialized enterprise agents. Both are selling custom agents grounded in company specific data and judged against task specific rubrics, but Mercor’s edge is its expert network and evaluator supply, while Snorkel’s edge is reusable data development and programmatic evaluation software.3 6 7

If this works, Mercor can climb from a fast growing labor marketplace into a much larger agent deployment layer inside enterprises. The long term prize is owning the loop between workflow mapping, expert evaluation, production monitoring, and data licensing, which would make Mercor useful both to operating companies trying to automate work and to AI labs seeking real enterprise environments.1 2 5 8