Job Description Drafting Agent · Use case
How can we use an AI agent to draft job descriptions from a hiring manager's brief on our own infrastructure?
A job description drafting agent writes a first draft of a job description from the hiring manager's brief and the company's role profile, in the house style, and marks wording that could put off or exclude a group of applicants, with the reason. The recruiter and the hiring manager edit and approve the text. The agent posts nothing, reads no application and no candidate's data, and sets no requirement the brief does not state.
The brief is three lines.
The last advert is the template.
- ✗ 01A hiring manager sends a short brief, and the job description is built by copying the last advert for a similar role, with its old requirements still in it.
- ✗ 02A requirement or a phrase that the role does not need can keep people from applying, and nobody rereads the text for that before it goes out.
Where it stops for a person
Always. The recruiter and the hiring manager approve before anything is posted.Always. It drafts and marks, it does not decide what the role needs: every requirement is the hiring manager's, and every marked phrase is a question, not a ruling. A draft with nothing marked has not been cleared: the marks show your guidance and nothing more. Pay is stated only as the approved band, and the agent proposes none.
It drafts and marks the wording.
The recruiter and the hiring manager approve.
Step · 01
A recruiter uploads the hiring manager's brief and names the role.
Step · 02
The agent reads the role profile, the grade and the style guide from the knowledge source.
Step · 03
It drafts the job description in the house format: the purpose of the role, the duties, and the requirements the brief and the profile state.
Step · 04
It marks wording that could put off or exclude a group: a requirement the brief does not support, an age-coded or gender-coded phrase, a length of experience given with no reason, with the passage and the reason.
Step · 05
The recruiter and the hiring manager edit the draft, decide every marked point and approve the text. A person posts it.
Reads from · connectors
The hiring manager's brief, as an uploaded file; and role profiles, grade descriptions, the style guide and your guidance on inclusive wording, indexed as a knowledge source with each reader's permissions kept. No application, no candidate and no employee record.
Acts through · MCP servers
Nothing. It has no tool that posts a vacancy, changes a role profile or sends anything.
What you set up
- Index the role profiles, grade descriptions, the style guide and your guidance on wording as a knowledge source, and remove the versions no longer in force.
- Give the agent the house format for a job description.
- Build the agent in Katonic Studio and test it on briefs for roles you have already advertised.
- Publish it to the recruiting team, one job family first.
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. Briefs, role profiles and draft job descriptions stay inside the company.
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 it post the vacancy?
No. It has no tool that posts or sends anything. The recruiter and the hiring manager approve the text, and a person posts it.
Does it read applications or candidates' data?
No. It reads the brief, the role profile and your guidance. It reads no application and no record of any person.
If nothing is marked, is the wording cleared?
No. The marks show where the draft departs from your own guidance. The recruiter and the hiring manager still read and approve every job description.