"Pay: $15 to $22 per hour. Full-time, 40 hours per week. Requirements: 2+ years of supervisory experience."
That's the job post Tamsin, owner of Puddlewick Bakehouse (a made-up neighborhood bakery), got back when she asked an AI to "write a job post for a counter lead."
It read well. Almost none of it was her job.
A job post is a promise. Make the AI promise only what you wrote down.
AI Prompts, Reviewed takes one prompt or workflow from How I AI, Claire Vo's show where guests share their screens and walk through how they really use AI. We run it on a small-business job, score it, and tell you what we'd change.
Credit: Based on "How to Standardize Job Descriptions Across a Company Using Claude Projects," a workflow from How I AI (hosted by Claire Vo, written up by the ChatPRD team), shown by Zach Leach, Head of Design at Gamma, in the episode "How I AI: Gamma's 3-Step AI Workflow for Global Feedback, Art Direction, and Hiring." Original workflow: https://www.chatprd.ai/how-i-ai/workflows/how-to-standardize-job-descriptions-across-a-company-using-claude-projects. Full episode: https://www.chatprd.ai/how-i-ai/gammas-ai-for-design-art-direction-hiring. Guest site: none listed on the How I AI workflow page.
What the original does
Gamma wanted hiring managers to stop writing job posts from scratch. The workflow has 4 steps:
- Set up a Claude Project loaded with strong past job posts plus company voice, pay, location and legal guidance.
- Add an instruction telling Claude which parts are fixed and which must come from the hiring brief.
- Give it a structured role brief and ask for a draft.
- Have the hiring manager and the people who own pay and legal review it before it goes live.
The line we'd tape to the monitor: "Return MISSING_INFORMATION for unsupported level, scope, requirements, benefits, or process details."
The test
Everything here is made up. Puddlewick is a bakery with a counter team of 3. Tamsin needs a counter lead. Her real job:
- Pay of $17 to $19 an hour
- Four shifts of 7 hours, Wednesday to Saturday, 6:00 a.m. to 1:00 p.m., for 28 hours a week
- Opens the shop, runs the register, trains new counter staff
- No supervisor experience needed, since she trains leads herself
Run 1, the shortcut: "Write a job post for a counter lead at my bakery."
The AI filled every gap with a guess. It posted $15 to $22 an hour, full-time at 40 hours, and asked for 2+ years of supervisory experience and a "degree in hospitality preferred."
Do the math on the top of each range. The post suggested up to $880 a week ($22 x 40). The real job tops out at $532 ($19 x 28). That's a $348 a week gap.
In one week, 31 people applied. At the phone screen, 9 dropped out once they heard the real pay and hours. At about 20 minutes a call, that's 180 minutes, or 3 hours of Tamsin's time on people the post had misled.
The rerun
Run 2: same AI, but Tamsin gave it a short brief with her pay range, hours, duties and must-haves, and told it to list anything missing before writing a word.
It came back with a MISSING list first: start date, how tips are split, and who's in charge on her days off. She answered in two minutes.
The post said $17 to $19 an hour, 28 hours, Wednesday to Saturday. Requirements were things she actually checks: early mornings, cash handling, and comfort training new staff. "Supervisory experience" became a preference she could teach.
Nobody dropped out at the phone screen over pay. The ones who called already knew it.
Verdict: 8/10
Strengths: the MISSING_INFORMATION rule stops the AI from inventing pay, hours and requirements. Fixed blocks keep every post consistent. And a human reviews it before it goes live.
Weakness: it's built for a company with a pay policy, a legal template and several reviewers. A one-location owner has none of those written down, so the setup step looks bigger than it is. And the episode's live demo used a one-line prompt for a joke role, which is exactly the shortcut that burned Tamsin.
Founder tweak: skip the big project setup. Write a 6-line brief (pay, hours, schedule, duties, must-haves, can-teach), paste it in every time, and make the AI ask before it fills a gap.
Copy the MCAFax version
This is our own rewrite, not the original. Copy the original from Zach Leach on How I AI: https://www.chatprd.ai/how-i-ai/workflows/how-to-standardize-job-descriptions-across-a-company-using-claude-projects
You write job posts for [business name], a [type of business] in [city, state].
Use ONLY the facts in my brief. Never guess pay, hours, schedule, benefits, or requirements.
My brief:
Role: [job title]
Pay: [exact range, per hour or per year]
Hours and schedule: [hours per week, days, shift times]
Duties: [3 to 6 things this person does every week]
Must-haves: [only what I will actually check]
Can-teach: [skills I am willing to train]
How to apply: [email, link, or walk-in]
Step 1. List anything you need that my brief doesn't give. Label it MISSING. Stop and wait for my answers.
Step 2. Write the post in this order: the job, a typical day, pay and schedule, must-haves, nice-to-haves, how to apply. Copy pay and hours exactly as I wrote them.
Step 3. Under the post, list each requirement you used and the brief line it came from. Flag any that came from nowhere.
Use plain, friendly words. Don't ask for a degree or years of experience unless my brief does.
Paste into ChatGPT, Claude, or Gemini. Save your brief so the next post is quick.
Before you post it
- Check the pay and hours line against your brief, word for word.
- Delete any requirement you won't actually check in an interview.
- Move "nice to have" skills you can teach out of the must-haves.
- Look up whether your state or city requires a pay range in job posts.
- Screen the applicants with a matching rubric: https://mcafax.com/notebook/prompts/job-application-screening
More owner prompts: https://mcafax.com/notebook/prompts