Recruiting Agent · Use case
How can AI screen candidates in Greenhouse and leave the decision to a recruiter?
A recruiting agent reads applications in Greenhouse or Lever, compares each candidate with the criteria written for the role, and explains where the candidate matches and where not. It recommends and a recruiter decides: the agent does not reject, advance or contact anyone. It runs on your own infrastructure, so candidate data is not sent to a public AI service.
Every application, read by hand.
And decisions that need a reason.
- ✗ 01Recruiters read every application against the same criteria, and good candidates wait while they do.
- ✗ 02Candidate data is personal data, and a screening decision made by software with no explanation is hard to defend.
Where it stops for a person
Always, before any decision on a candidate.Always. Every recommendation goes to a recruiter, who makes the decision and records it in the applicant tracking system.
It recommends.
A recruiter decides.
Step · 01
The agent is connected to the applicant tracking system and given the criteria for the role.
Step · 02
It reads each application and CV.
Step · 03
It compares the candidate with each criterion and notes the evidence it found.
Step · 04
It gives the recruiter a summary and a recommendation, with its reasons.
Step · 05
The recruiter decides, and records the decision in the applicant tracking system.
Reads from · connectors
Greenhouse
Lever
The agent connects to Greenhouse or Lever through connectors in the Katonic platform.
Acts through · MCP servers
Nothing: it recommends and a person decides. It does not change a candidate's stage, schedule an interview or send a message.
What you set up
- Connect Greenhouse or Lever.
- Write the criteria for each role, and what counts as evidence.
- Review a sample of recommendations with recruiters before release.
Where it runs
The agent, its knowledge store and the model run on infrastructure you own or control. You choose the model and serve it yourself, so questions, documents and answers are not sent to a public AI service.
How you get it today
This page is labelled Use case: the agent is built in Katonic Studio. You describe the job, connect what is named above, add test questions and release the agent once it passes them. When the Catalog holds a tested template for this job, this page will carry the Template label and show how to take it.
Does the agent reject candidates?
No. It recommends, with reasons, and a recruiter decides. It cannot change a candidate's stage.
Can it schedule interviews?
No. It does not write to the applicant tracking system or to a calendar. A person schedules.
How does a recruiter check a recommendation?
Each recommendation lists the criteria and the evidence found for each, so the recruiter can read the reasoning against the application.
Where is candidate data processed?
On your own infrastructure, with a model you serve. Applications are not sent to a public AI service.