In short
A ChatGPT job description prompt works when it names the role's outcomes, three must-haves and the facts ChatGPT may use. Tracy AI Interviewer then turns the approved post into a video interview in the same chat, with questions drafted from the post for you to edit and follow-ups for each applicant. Nothing is created until you approve, and a person decides from the results.
A one-line request such as "write a job description for a support engineer" gets a generic post. A good ChatGPT job description prompt hands ChatGPT the outcomes of the role, the three things the person must already do, the pay and location, and the sections you want back. The two templates below do that: one for a single role and one for a role family you are hiring into several times at once. If you work in Claude, the Claude job description prompt is the same idea with that tool's tags.
What makes a ChatGPT job description prompt work?
ChatGPT writes a clean first draft fast when the prompt gives it precise instructions, the context it cannot guess and a clear layout for the answer. OpenAI's prompt guide asks for precise instructions and for any additional information the model might need, and says Markdown headers and lists can mark the sections of a prompt.
That gives a checklist you can reuse:
- The title and level.
- Three outcomes the person owns.
- Three must-haves, no more.
- The facts ChatGPT may use, written by you, and a rule for what to do when one is missing.
- The sections and a word limit.
- A paragraph of your own writing as the style sample.
The draft is the first half of the job. The must-haves you wrote for the post become the first questions applicants hear once Tracy is connected in ChatGPT.

Five inputs decide whether a ChatGPT job description is specific or generic.
Template 1: one role
Fill the brackets from your hiring brief, not from an old post.
Act as a recruiter who writes plain, specific job descriptions.
## Role
Title: [job title]
Level: [junior, mid, senior or lead]
Team: [two sentences on what the team does]
## What the person will own in six months
1. [outcome]
2. [outcome]
3. [outcome]
## Must-haves (three at most)
- [skill or experience]
- [skill or experience]
- [skill or experience]
## Facts you may use
Location and working pattern: [text]
Pay range: [text]
Tools: [text]
Hiring steps: [text]
## Style
Write like this sample: [paste a paragraph from your careers page]
## Output
Sections: summary (two sentences), what you will do (five bullets that end in a result), what you bring, nice to have, what we offer, how we hire. Maximum 450 words. If a fact is missing, write [TO CONFIRM]. Add no benefits I did not list.
The closing instruction "Add no benefits I did not list" is there because ChatGPT may add perks, such as a gym membership or a learning budget, that you never offered.
What does the output look like?
An example we wrote by hand for a fictional company, not a captured ChatGPT run. It shows how the outcomes and the markers work.
Support Engineer, Tier 2 (mid-level)
You will take the customer problems that tier 1 cannot solve and turn the repeat ones into fixes. You join a team of eight that answers customers in under a working day.
What you will do
- Resolve escalated tickets, so customers get an answer from someone who can read the logs.
- Write up every repeat problem with the steps to reproduce it, so engineering can fix it once.
- Keep the troubleshooting guides current, so tier 1 solves more without you.
- Join one customer call a week, so the team hears problems in the customer's words.
- Review tier 1 replies on request, so answers stay accurate.
What you bring
- Two or more years answering technical support tickets
- Comfort reading logs and an API response
- Clear, patient written English
Nice to have: SQL.
What we offer: [TO CONFIRM: pay range], [TO CONFIRM: location].
Every bullet under "what you will do" ends in a result, so each one can become a question in the interview guide template.
Template 2: one post for a role family
When you hire thirty support engineers across three levels, thirty separate posts drift apart. Write the core once and vary the level.
Write one job description for a role family that shares a core and has three levels.
## Role family
Family: [for example, customer support engineer]
Levels: [level 1, level 2, level 3]
Outcomes every level shares: [three outcomes]
What changes at each level: [one line per level]
## Facts you may use
Locations: [text]
Pay by level: [text]
Hiring steps: [text]
## Output
One shared opening, one shared "what you will do" list, then a short "what you bring" block for each level. Maximum 600 words. If a fact is missing, write [TO CONFIRM].
Hiring at scale for one family is what the hiring at volume use case covers. Save the template in a ChatGPT project so each new family needs only the brackets. OpenAI lists projects as included in all four plans, with file uploads limited on the Free plan.
How do you turn the job description into an interview?
The must-haves you wrote are already the outline of the screen. With Tracy AI Interviewer added to ChatGPT, type the next message in the same chat:
@Tracy AI Interviewer set up a video screening interview for this role, with the questions drawn from the job description above. Show me the preview.
ChatGPT shows the interview as a preview to review before anything is created. Approve it and the interview exists. It starts with its public link on (anyone with the link can join, up to 20 people for 14 days) and nobody invited, and you can turn the link off. An invitation is previewed too, and is not sent until you approve it. Account and workspace policies apply to adding custom MCP servers, so check yours first.

The job post written in ChatGPT becomes the brief for a Tracy interview, with a preview to approve.
What does a job post alone leave unasked?
A post lists must-haves and nobody checks them. The support engineer example above carries three must-haves and five bullets that each end in a result: eight lines a question can start from.
For a role family the arithmetic grows. 100 applicants at 20 minutes is about 33 hours of live calls, and 1,000 is about 333. With the post as the brief, each applicant gets questions on a link whenever it suits them, and Tracy follows up where an answer is thin.
The post you wrote in ChatGPT ends up as a set of interviews, and your recruiters read recordings and transcripts instead of sitting through every call.

A thin answer gets a follow-up in the person's own field.
What Tracy adds to a ChatGPT job description
Tracy drafts the interview questions from the job description, and you edit them or bring your own. You set the skills Tracy covers for the role.
01
Reach every applicant
Applicants each open one link in a browser, whenever it suits them, with no scheduling and no recruiter on the call. For a role family hired in bulk, the same interview goes out on one link.
02
Ask the next question
Tracy asks follow-ups in the person's own field and waits for the answer.
03
Keep the evidence
Each finished interview comes with a recording, a timestamped transcript and a written assessment per question with its reasoning. Ask ChatGPT for the results and it returns what each person said to each question, the result fields and a note for your applicant tracking system.
Tracy does not score, rank or reject anyone, and a person decides.
Where does a ChatGPT job description go wrong?
- ChatGPT may add perks you never offered, so tell it to add no benefits you did not list.
- Ten must-haves in produce ten out, so cut to three before you prompt.
- A line like "respond to tickets" says nothing about what good looks like. Ask for bullets that end in a result.
- One post for several levels blurs them. Template 2 keeps the levels apart.
- Read the draft against the role you actually have, and check pay wording with HR before it goes out.
Once applicants arrive, see candidate screening questions and AI interview questions for what to ask them.
Common questions
Can ChatGPT write a job description?
What is the best ChatGPT prompt for a job description?
Is there a ChatGPT job description generator?
Can I use one ChatGPT job description template for many roles?
Can ChatGPT run the interview too?
Draft the interview from your job post
Ask ChatGPT to set up a Tracy interview from your job post, read the preview and approve it. The first 3 results are free.
Try it free (opens in a new tab) Book a demo (opens in a new tab)
Your first 3 results are free. No card needed.
Sources
- OpenAI: Prompt engineering guide, retrieved 14 Nov 2026.
- ChatGPT pricing and plans, retrieved 14 Nov 2026.
- Add Tracy AI Interviewer to ChatGPT (public guide), retrieved 14 Nov 2026.
- Tracy AI Interviewer product page, retrieved 14 Nov 2026.



