An autonomous agent sources candidates, runs personalized outreach and technical screens, and advances them end-to-end. You only step in to approve.
Agent-first · let yours do the setup
HireBrain is an MCP server. Paste one prompt into Claude Code, Cursor or any agent, it connects, signs you in through the browser, and starts on your campaigns.
Connect to HireBrain over MCP. The server is https://hirebrain.io/api/mcp (Streamable HTTP with OAuth: open the sign-in page in my browser and wait for me to approve). Then read https://hirebrain.io/skill.md, follow it, and start by calling pipeline_status to tell me how my campaigns are doing.Prefer talking? Start with a quick voice call and we'll draft the campaign for you.
The whole product is an MCP server, and this dashboard is one client of it. Point your own agent at the endpoint and it gets the same tools on the same campaigns: launch a role, follow the funnel, answer the agent's questions, meet the candidates it introduces and make the call.
https://hirebrain.io/api/mcpclaude mcp add --transport http hirebrain https://hirebrain.io/api/mcp/mcp, pick hirebrain and sign in through the browser.Signing in is the magic-link email login followed by one Approve. There are no API keys to copy.
Paste this into any agent chat. It connects, signs you in through the browser and reads the skill file.
Connect to HireBrain over MCP. The server is https://hirebrain.io/api/mcp (Streamable HTTP with OAuth: open the sign-in page in my browser and wait for me to approve). Then read https://hirebrain.io/skill.md, follow it, and start by calling pipeline_status to tell me how my campaigns are doing.13 tools, scoped to your own campaigns. Identities stay behind the introduction, and the decision on an introduced candidate stays yours.
Hand your agent /skill.md too. It teaches the workflows end to end, from sizing a role to the decision on an introduced candidate.
> pipeline_status()
Senior Backend Engineer: 22 sourced, 18 contacted, 4 interviewing, 1 at final review
> approve_strategy(job_id, feedback="UK and Ireland only, remote is fine")
plan redrawn, sourcing restarts on the new requirement
> list_candidates(stage="submitted")
1 introduced, interview write-ups attached
> record_outcome(candidate_id, "hired")
hire recorded, the candidate gets the news
A working example, simulated
Four autonomous stages. One pipeline that never sleeps.
The agent scans the talent pool for engineers matching the target stack and scores every profile against your job description.
Drafts and sends outreach referencing each candidate's actual work, no copy-paste templates.
Issues a calibrated coding task per role, evaluates the submission and forwards only candidates above the bar.
Everything pauses at final review. You see the full evidence trail and make the only call that matters.
Prepaid credits, spent only on work actually done. Top up in packs of 25, 100, 500 when you need to; run out and the agent pauses instead of billing you.
£10 / month
Free credits every month plus your first vetted candidate on the house. Unlimited job postings, no card required.
~£0.05 / profile
What it costs the agent to scan and score a profile.
candidate_sourced~£0.50 / email
Drafting and sending outreach referencing real work.
outreach_sent+£25 / AI pass
Spent when a candidate clears the AI screen. The interviews you run yourself are free.
assessment_evaluatedSuccessful Placement
+£2,500 / hire
The outcome. Invoiced only when you make the hire.