By PitchMe has launched a remote Model Context Protocol (MCP) server that lets recruiters run candidate searches through ChatGPT, Groq, and Muse, according to HR Tech Feed. The launch was announced November 9, 2026, from London and Arlington, Texas.
Recruiters can describe a role in plain language, ask the assistant to search PitchMe, build a shortlist, and reveal available contact details within their own PitchMe account. Shortlists can be sent to Bullhorn or Airtable. The MCP server is live now for PitchMe customers, with setup requirements varying by assistant.
What the connector actually does
The workflow starts with a prompt. HR Tech Feed gives the example: “Find ICU nurses in Texas with active state licenses, shortlist the top 20 and reveal their contacts.” The assistant then queries PitchMe’s candidate data, applies the recruiter’s stated criteria, and returns a shortlist.
Dina Bay, founder and CEO of PitchMe, framed the move as removing friction. “Recruiters don’t need another tab. They need the tools they already use to do more of the work,” she said.
The shortlist is not just a list of names. Contact details are revealed inside the recruiter’s own PitchMe account, which means access controls and visibility rules remain tied to the agency’s existing permissions. Max Kuzkin, PitchMe’s CTO, said the agent performs actions on the recruiter’s behalf and that work is visible inside PitchMe. “Recruiters can inspect the search criteria, refine their instructions and see the project and shortlist as they develop,” he said.
Where this fits in PitchMe’s product line
PitchMe describes itself as the data intelligence layer for recruitment and staffing agencies. The company serves agencies across the UK, US, and Europe, and recently expanded search coverage in Belgium, Bulgaria, Luxembourg, and the Netherlands.
In September, PitchMe launched SONAR at Staffing Industry Analysts’ CollaborationX conference in Dallas. SONAR is a unified search across an agency’s own database and the external talent market. The MCP server extends that capability into third-party AI assistants rather than requiring recruiters to work inside PitchMe’s own interface.
For healthcare staffing agencies, the data layer includes the US national provider dataset, NPPES, with matched candidate records carrying NPI numbers. That matters for roles where license verification is part of the shortlisting process.
What this means for HR and hiring teams
The practical shift is that candidate search becomes conversational. A recruiter or HR manager can describe a role in the same terms they would use in a briefing call, and the assistant translates that into structured search criteria.
The integration with Bullhorn and Airtable means the shortlist lands in the system where the rest of the hiring workflow already lives. That reduces the manual step of exporting and importing candidate data between tools.
Bay made the underlying point directly: “An AI assistant is only as useful as the data underneath it. By connecting AI assistants to verified candidate data, we’re turning a prompt into a shortlist that recruiters can actually call, already filed in their ATS, with the agency in control of who can see what.”
For payroll and finance leads, the relevant angle is time. If sourcing and shortlisting moves from hours of manual searching to minutes of prompt refinement, the cost per shortlist falls. But no pricing or fee figures were provided in the announcement, so teams should ask PitchMe directly about commercial terms.
Setup and access considerations
Setup requirements vary by assistant. PitchMe has published setup guides for the MCP server, and agencies can request a demonstration through the company’s site.
Two compliance points are stated in the announcement. PitchMe is SOC 2 certified and GDPR compliant. For agencies handling healthcare candidate data or operating across European markets, those certifications address part of the due diligence checklist. However, readers should confirm how the MCP connection handles data in transit and what logs are retained, since the announcement does not detail those specifics.
Agencies using Avionte are also named in the announcement, though the shortlist export examples given are Bullhorn and Airtable. Teams on other ATS platforms should confirm whether their system is supported before assuming the workflow will connect.
What remains open
The announcement does not state whether the MCP server supports all PitchMe search features or a subset. It also does not specify whether the assistant can update candidate records or only read and shortlist them. Kuzkin’s comment about the agent performing actions on the recruiter’s behalf suggests some write capability, but the scope is not detailed.
For international teams, the expansion of search coverage in Belgium, Bulgaria, Luxembourg, and the Netherlands is recent. Recruiters should verify that the markets they hire in are covered before relying on the connector for those searches.
Sources: HR Tech Feed
Photo: Rent-A-Hitman Recruitment & Elimination Department by ChatGPT, Public domain
