When you recruit on Get on Board with an AI agent, two layers of signals work in sequence. Before anyone applies, your agent shortlists candidates using the profile fields the MCP server returns — the recruiter summary, skills, seniority, country, and English level. Once candidates apply to your job, Superpower AI takes over inside Get on Board with match percentages and screening. The agent handles discovery; Superpower handles evaluation.
Where each signal lives
| Signal | Where you see it | When it applies |
|---|---|---|
| Recruiter summary and profile fields | MCP results in your agent | While sourcing, before anyone applies |
| Match percentage | Your job’s application list in Get on Board | After a candidate applies to a job with Superpower enabled |
| Superpower AI screening | Your hiring process in Get on Board | While reviewing and shortlisting applicants |
The match percentage is not part of MCP results. It compares an application against one specific job, so it can only exist once someone has applied. MCP results describe candidates who have not applied to anything yet — your agent builds its own ranking from the fields it receives.
What the MCP fields tell your agent
Every search_talent result includes a set of privacy-safe fields your agent can reason with:
- Recruiter summary: a short, AI-written overview of the candidate’s experience, generated from their profile without contact details or personal identifiers. It is the densest signal in the result — your agent can compare it directly against your role’s requirements.
- Skills and tags: the candidate’s top technologies and areas.
- Seniority and years of experience: when available on the profile.
- Country and English level: useful for location and language requirements.
Because these fields travel with every result, you can ask your agent to post-filter and rank: “keep only senior profiles with React and advanced English, then order them by relevance to this job description”. If you already posted the role on Get on Board, your agent can read the full description with job_details and use it as the comparison brief. That ranking is the agent’s own judgment over the returned fields — useful for a shortlist, but not the same number as the in-app match percentage.
Where the match percentage comes in
The match percentage appears on each application when your job has Superpower AI enabled. It compares the job description with the CV text, the application answers, and, when the candidate links one, a public GitHub profile. That is richer input than any sourcing signal: it includes documents and answers that exist only after someone applies.
So the two numbers answer different questions. Your agent’s ranking answers “who looks worth contacting?”. The match percentage answers “how close is this actual application to this job?”. Treat both as signals to prioritize review, never as the final decision.
There is also a signal layer only a marketplace can hold: because professionals apply to many companies through Get on Board, the platform sees how profiles fare across separate hiring processes — a cross-company record that no single-company ATS or do-it-yourself agent ever sees.
Where Superpower AI picks up after the shortlist
An agentic shortlist ends where your hiring process begins. From the MCP results, you open each candidate’s profile_url in Get on Board, where you can unlock contact details with your Talent Database credits and invite the person to a process. Unlocking and inviting happen in the web app — the MCP server is read-only and never returns contact data.
Once candidates enter your process, Superpower AI adds the evaluation layer:
- A match percentage on each application, with color cues in List view.
- AI-generated summaries and screening support to review and shortlist applicants.
- The Selected by Superpower phase, which pre-groups the strongest applications for you.
A typical flow, end to end
- Your agent reads your job with
job_details, or you give it a brief in plain text. - It searches Talent Database with
search_talentand ranks results using the recruiter summary, skills, seniority, and English level. - You review the shortlist and open the profile links in Get on Board.
- In the web app, you unlock the candidates you want to contact and invite them to your process.
- As applications arrive, Superpower AI scores each one with a match percentage and helps you identify the best-matching applicants.
Each layer feeds the next, and the data stays under Get on Board’s rules at every step: the agent only ever sees privacy-safe summaries, and the evaluation of real applications stays inside your account.
Related articles
- What is the Get on Board MCP Server?
- How is the match percentage between a candidate and a job calculated?
- What is Superpower AI and what does it include?
- How to use Superpower AI to review and shortlist applicants
- How can I use Get on Board for agentic recruitment?
- How does the MCP server protect candidate data?
More from Products and add-ons
Sign in with Google
