Symbory speaks MCP and REST. The assistant your team already uses can ask about a person's history and get a real answer, with the same permissions the person asking already has. We're not another place to go and look. We're where the answer comes from.
GET /v1/timeline/employees/{employeeID} Authorization: Bearer sk_live_…
200 OK [ { "type": "interview_completed", "occurred_at": "2024-02-14T10:00:00Z", "body": "Wants architectural ownership." }, { "type": "commitment_logged", "occurred_at": "2024-03-01T09:15:00Z", "body": "Salary review at twelve months." }, { "type": "oneonone_completed", "occurred_at": "2025-02-14T11:00:00Z", "body": "Review held. Scope widened." } ]
What an assistant reads before someone's 1:1. The first two events pre-date the hire: the query reaches back through candidate_id, so one call returns the whole line.
Not because we built a connector for it. Because the whole product, ATS included, is reachable over an open REST API and an MCP server, and you write the twenty lines that fit your stack. Nothing waits on our roadmap.
Symbory ships a Model Context Protocol server. Point Claude, or any MCP-capable assistant or internal agent you're building, at a workspace and it can ask questions against a person's real history: hiring conversations, one-on-ones, notes, commitments, goals.
recall_commitment answers a plain-language question against everything a person's history holds. This is the one nothing else does: every other tool follows what the employee delivered. This one holds the company to its half.
prepare_1on1 builds a briefing from the whole relationship, not the last meeting. get_employee_timeline returns the run of it: first interview through today, one continuous line. search_memory looks across all of it at once.
Tools run through the same authorisation as the product itself. The token carries a role, scoped to your workspace. An assistant can't reach anything the person holding the token couldn't already open. Nothing is a direct line to the data.
Fourteen tools today, covering people, hiring and the timeline. Want to see it against your own setup? Get in touch.
Opening people data to an AI assistant is a reasonable thing to be careful about. Here is exactly where the edges are.
No, it inherits them. Every tool call runs through the same authorisation and workspace scoping as the product, and the token carries the role of whoever generated it. There is no path from a tool straight to the underlying data.
One workspace, one role, one year. A token carries the role of whoever generated it and can do nothing that person couldn't, so it is worth issuing from an account with the access the job actually needs. It's shown once, so keep it somewhere you'd keep any other secret. Full GDPR rights apply to everything behind it: access, correction, deletion.
The dashboard is a client of this API, not a privileged path around it. Mint a token with POST /v1/auth/api-token from Settings, on the Team plan and up, valid a year, shown once, carrying your own role. No sales call.
/v1/talent/jobsList and open roles/v1/talent/jobs/{jobID}A single role/v1/talent/jobs/{jobID}/applicationsEveryone who applied/v1/talent/jobs/{jobID}/applications/{applicationID}/stageMove someone along the pipeline/v1/talent/candidatesThe people, across every role/v1/talent/candidates/{candidateID}One candidate/v1/talent/candidates/{candidateID}/applicationsEvery role they've been up for/v1/talent/candidates/{candidateID}/documentsCVs and attachments/v1/talent/applications/{applicationID}/notesWhat the team wrote down/v1/talent/applications/{applicationID}/stage-historyEvery move, with timestamps/v1/talent/applications/{applicationID}/rejectClose it out, in their language/v1/hiring/interviewsWhat's scheduled/v1/hiring/applications/{applicationID}/interviewsBook against an application/v1/hiring/interviews/{interviewID}One conversation/v1/hiring/interviews/{interviewID}/completeClose it and write it back/v1/people/employeesEveryone currently employed/v1/people/employees/{employeeID}One person/v1/people/employees/{employeeID}/documentsContracts and acknowledgements/v1/people/employees/{employeeID}/one-on-onesThe conversations themselves/v1/growth/goalsAcross the company/v1/growth/employees/{employeeID}/goalsWhat one person agreed to/v1/growth/goals/{goalID}A single goal and its progress/v1/timeline/employees/{employeeID}First interview through today, one continuous line. The call the MCP tools sit on/v1/public/tenants/{tenantSlug}/jobsOpen roles, no auth. Build your own careers page/v1/public/applicationsApplications land straight in the pipeline/v1/talent/applications/{applicationID}/summaryA CV read for you/v1/talent/applications/{applicationID}/interview-prepWhat to ask this person/v1/people/employees/{employeeID}/askA plain question against one person's history/v1/people/employees/{employeeID}/one-on-ones/briefingWhat's open, and what was promisedThat's the set most integrations start with. The rest of hiring, people and growth follows the same shape and your token already reaches it. If you'd rather someone walked it through against your stack, get in touch.
Set up a workspace in a minute, generate a token in Settings, and make the first request against real data. Nothing is charged today, and we'll migrate what you already have.
Create your account →Rather talk to a person first? Get in touch.