How to Rewrite an AI Job Description Before You Post It

An AI-drafted job description usually reads fine and says almost nothing. Every line could belong to any company, so a good candidate can't tell what the job is, and a weak one applies anyway. If your company allows AI for job ads, the draft is a starting point. The rewrite is where the ad starts to describe your actual role.

This page walks through a workflow for recruiters and hiring managers, shows a before and after, and covers common objections.

What goes wrong in AI-written job descriptions

The problems are consistent, whichever model produced the draft:

• Stacked adjectives. "Dynamic," "passionate," "fast-paced," "innovative." None of them tell a candidate what they'll do on Tuesday.
• Verbs with no object. "Drive excellence," "foster collaboration," "play a pivotal role." Drive what? Collaborate with whom?
• Invented requirements. Models add "excellent communication skills" and "a proven track record" because job ads usually contain them, not because your team needs them.
• Missing facts. The model doesn't know your pay band, your tools, your manager, or why the last person left. So the ad leaves those out.
• Even rhythm. Every sentence is about the same length and every bullet starts the same way. It feels like a template because it is one.

We keep a longer list in words that make writing sound like AI. Most of them show up in job ads within the first paragraph.

A workflow for rewriting an AI job description

1. Gather the facts before you edit

Write down what the model couldn't know: the team size, who the role reports to, the three tools used daily, the pay range, the location or remote policy, and the first real project. Ten minutes with the hiring manager gets you most of it. Ask one question: "What will this person have done by day 60 that makes you glad you hired them?"

2. Run the draft through a checker

Paste the draft into Plainspoke and see which phrases read as generic or machine-made. Treat the result as a list of places to look, not a verdict. Every flagged line is a question: is there a specific fact that belongs here instead?

3. Replace abstractions with specifics

For each flagged sentence, swap in a task, a number, a tool, or a name. If you can't find a fact to swap in, cut the sentence. A shorter ad with real details beats a longer one with filler.

4. Trim the requirements

Split requirements into "needed on day one" and "we'll teach you." Anything you can't defend in an interview doesn't belong in the first list.

5. Have the hiring manager read it

Ask them one thing: "Is this the job?" They will catch what a checker can't, like a tool you no longer use or a responsibility that moved to another team.

6. Read it aloud, then post

If you'd feel odd saying a sentence to a candidate across a table, rewrite it. Then run a last check and publish.

Worked example: before and after

The example below is made up for illustration. The company, tools, and pay figures are invented. The point is the kind of detail you'd supply from your own role.

Before (AI draft)

We are looking for a dynamic and passionate Operations Analyst to join our fast-paced, innovative team. In this exciting role, you will play a pivotal part in driving operational excellence across the organization. The ideal candidate is a self-starter with excellent communication skills and a proven track record of success. We offer a competitive salary and a collaborative culture where your contributions will make a real impact.

After (rewritten with facts from the hiring manager)

Operations Analyst, Dispatch (hybrid, Columbus)

When a delivery route falls apart, dispatchers message you. Most weeks you'll pull the last two days of delivery data, work out why a dozen trucks ran late, and tell the dispatch leads what to change before Thursday's planning call.

What you'll do

• Own the weekly on-time-delivery report (SQL and Looker).
• Find patterns in late deliveries and write them up in a page or less.
• Work with two dispatch leads and the customer support manager, who will send you most of your questions.

What you need

• Two or more years in analytics or operations.
• You can write a SQL join without help.
• You can explain a number to someone who doesn't like numbers.

Pay and setup: $68,000 to $78,000. Three days a week in our Columbus office, two at home.

What changed

The rewrite is not more polished. It's more specific, and it's a little rougher in places on purpose. Only the hiring manager could have supplied most of those details, which is the reason a candidate can believe them.

Objections, answered straight

"If our company approves AI for job ads, why rewrite at all?"

Approval covers whether you may use AI. It doesn't make the output good. A generic ad costs you applications from people who didn't understand the job, and it costs you the ones who skimmed it and moved on.

"Candidates can't tell, and don't care who wrote it."

They may not care who wrote it. They do notice when an ad says nothing. We don't have survey data on how candidates feel about AI-written ads, so we won't quote any. What you can check yourself is whether a stranger could tell your ad apart from ten others in the same field.

"Will this make my ad pass an AI detector?"

We can't promise that, and nobody honestly can. Detectors disagree with each other and sometimes flag text people wrote by hand, as we explain in why AI detectors flag human writing. The goal here is a better ad, not a score.

"It's faster to prompt again."

Re-prompting gives you a different generic draft. The slow part of a good ad is collecting the facts, and no prompt can do that for you. Once you have them, editing takes less time than most people expect.

"What about legal requirements like pay ranges?"

Rules on pay disclosure, protected characteristics, and automated tools vary by place and change. Plainspoke is a writing tool. It doesn't post your ad, audit it for bias, or check it against employment law. Have HR or counsel review the final version where it matters.

Where Plainspoke fits, and where it doesn't

Plainspoke checks and rewrites the language of professional writing, including business writing like job ads. It does not hold your applicant data, run your hiring process, or replace the hiring manager's knowledge of the job. If you also write outreach to candidates, see our pages for LinkedIn and cold email.

For how we think about making AI drafts sound like a person wrote them, read how to make AI writing sound human. When you're ready to try it on a real ad, sign up or check pricing first.

Check a draft free