ATS for humans and agents

A complete ATS — use the software, or use an agent.

Yardstick is a full applicant tracking system with two first-class ways to work. Want a software interface? Click through roles, candidates, interviews, and scorecards, and the app guides you the whole way. Prefer to delegate? Hand the same work to your agent — Claude Code, Codex, or an MCP-connected assistant. It is one system of record underneath, so you switch whenever it suits you, and you keep the wheel on the decisions that matter.

Book a callView pricing

Two interfaces

A software interface, and an agent interface — same system either way.

Yardstick is one system of record with two first-class interfaces. If you want a software interface, open the web app and manage roles, candidates, interviews, and scorecards yourself — it walks you through each step. If you want an agent interface, stay in the tool you already work from and let your agent operate the same data — Yardstick is agent-operable on every account through the yardstick CLI, the public API, and an MCP server. The operator can be a coding agent like Claude Code or Codex, or any MCP-connected assistant like ChatGPT or Claude.

That is the real “humans and agents” story: not an agent-only tool, and not a person supervising a bot, but one hiring system that is equally at home as software you click or an agent you delegate to. Nothing lives in a second place — the web app and your agent read and write the same records.

Agent interface

Delegate the whole hiring motion, when you want to.

Everything here you can also do yourself in the web app. When you would rather delegate, these are four motions your agent runs through the yardstick CLI, public API, or MCP — then hands you the decision.

  • Requisitions

    Ask your agent to draft a job description and a job-specific interview plan from the role you describe. It prepares the requisition; you approve it before it publishes.

  • Candidates

    Import candidates from your existing ATS with a preview-first dry-run, move people between stages, and summarize the evidence on a candidate. The agent prepares the change; you approve advancing or rejecting anyone.

  • Interviews

    Build interview plans, assign interviewers, and generate job-specific questions and rubrics, so every interview is set up to produce usable evidence.

  • Scorecards

    Generate a structured scorecard for the role and chase the missing ones with reminders to the interviewers who owe feedback.

You decide

The agent prepares the move. You approve it.

Delegating is only useful if you stay in control of the calls that matter. Your agent can draft, import, summarize, and prepare, but publishing a job, advancing or rejecting a candidate, and activating outreach that emails candidates are decisions a human approves. The agent shows you what will change and waits for your go — and the change is saved with an audit trail.

Codex

You

Move Morgan Reed to Screening, assign Alex as interviewer, and add a 30-minute screening task.

Codex

Prepared in Yardstick. Here's the change — approve and I'll save it.

- Stage: Applied → Screening

- Interviewer: Alex Carter

- Screening task added

Waiting for your approval before saving the candidate stage change.

Structured evidence

The agent is only as good as the system it drives.

Behind every agent action is a structured-interview hiring system, not a thin wrapper. Yardstick connects role requirements, candidate stages, interview plans, interviewer assignments, questions, rubrics, structured scorecards, and decisions — so the work your agent prepares lands in one place and produces comparable evidence.

That system can start before a role is even open: Yardstick's talent CRM builds prospects, pools, and human-approved outreach sequences into the same interview-led pipeline. And it starts with deciding what kind of hire the role actually needs — clone a strength or cover a gap — before a single requirement is written down.

Yardstick is a strong fit when

  • you run work from a coding agent or an MCP-connected assistant and want hiring in the same window
  • you want one system of record for roles, candidates, interviews, and scorecards
  • interview quality matters as much as pipeline tracking
  • you want agents to move fast on the operational work while you keep the decisions

FAQ

Common questions.

Can I run my whole ATS from my agent?

Yes. Yardstick is agent-operable on every account through the yardstick CLI, the public API, and an MCP server. Your agent can draft requisitions, import and move candidates, build interview plans and questions, and generate scorecards. The decisions that matter — publishing a job, advancing or rejecting a candidate, and sending candidate outreach — are approved by a human before they happen.

Which agents or tools can operate Yardstick?

A coding agent like Claude Code or Codex running the yardstick CLI, anything built on the public API, or any MCP-connected assistant like ChatGPT or Claude. You operate Yardstick through your own agent — it is not a Yardstick chatbot.

Does the agent make hiring decisions?

No. The agent prepares and explains the work and shows you what will change, but a human approves the decisions that matter: publishing a job, advancing or rejecting a candidate, and activating an outreach sequence that emails candidates. Every candidate email goes out through a sequence a human has approved.

How is this different from a generic ATS or an AI question generator?

A generic ATS tracks candidates and stages; an AI question generator makes questions in isolation. Yardstick connects the whole motion — roles, candidates, interview plans, rubrics, scorecards, and decisions — into one system of record you can operate yourself or drive from your agent.