- How to Prompt AI Tools for Job Descriptions - March 17, 2026
- 8 Best Job Description Builder Tools in 2026 - March 12, 2026
- Are DEI Initiatives Effective? Navigating a Transformed Landscape in 2026 - February 12, 2026
AI tools like ChatGPT, Claude, and Gemini can generate a job description in seconds. But for enterprise talent acquisition teams, “fast” is not enough. You need JDs that are accurate, inclusive, legally compliant, and on-brand at scale.
The difference between a bland, AI-generated JD and a great one almost always comes down to the prompt.
This guide provides the exact prompting techniques and ready-to-use templates for your TA team to get consistent and high-quality job descriptions from any AI tool.
Why Prompting Matters More Than the Tool
Most AI tools draw from the same underlying language model technology. A vague prompt produces a generic JD regardless of which tool you use.
A well-structured prompt that gives the model context, constraints, and clear output requirements consistently delivers better results.
Think of prompting as writing a brief for a skilled copywriter. The more specific your instructions, the less editing you’ll need later on.
The 6 Elements of a Strong JD Prompt
Every high-performing prompt for job description writing shares these six components:
| Element | What to Include |
| 1. Role Context | Job title, department, seniority level, and team structure |
| 2. Company Voice | Tone, values, and any brand language guidelines |
| 3. Output Format | Sections you need (e.g., About Us, Responsibilities, Requirements, Benefits) |
| 4. Inclusion Constraints | Instructions to use gender-neutral language, avoid jargon, limit requirements |
| 5. Length & Style | Target word count, bullet vs. prose, reading level |
| 6. ATS Optimization | List of 5–10 high-volume keywords or skills to ensure the JD is machine-readable and SEO-friendly |
Prompt Templates You Can Use Today
The following templates are tool-agnostic, meaning they work in ChatGPT, Claude, Gemini, or any other AI writing assistant. Copy, customize, and save them as your team’s standard starting points.
Template 1: Generate a JD from Scratch
The biggest mistake TA teams make when prompting AI for job descriptions is skipping the context step. They type in a job title, hit generate, and wonder why the output sounds like every other JD on the internet.
Before you open your AI tool, gather your inputs first. That means company-specific details but also the nuances from your hiring manager intake: the real requirements, the team dynamics, the success metrics, and the non-negotiables that never make it into a standard job description.
When you feed the AI that context upfront, you get a JD that’s right the first time. No back-and-forth email chain. No hiring manager saying ‘this isn’t what I asked for.’ Because it’s built from their actual words, not the AI’s best guess.
| PROMPT — Generate JD from Scratch |
| You are an expert HR copywriter for an enterprise company. I will provide raw notes from my intake meeting with the hiring manager. Use these notes as the sole source of truth for the responsibilities and requirements. Do not add standard industry responsibilities unless they are mentioned in the notes.[PASTE RAW NOTES/INTAKE MEETING TRANSCRIPT HERE]Using only that context, write a job description for the following role:Job Title: [INSERT TITLE] Department: [INSERT DEPARTMENT] Seniority: [e.g., Senior, Manager, Director] Location / Work Type: [Remote / Hybrid / On-site, City] Reports To: [INSERT ROLE]Format the JD with these sections:About [Company Name] (2-3 sentences)About the Role (3-4 sentences)What You’ll Do (6-8 bullet points)What We’re Looking For (5-7 bullet points, separate required from preferred)What We Offer (4-5 bullet points)Equal Opportunity StatementUse gender-neutral language. Avoid jargon. Keep the reading level at Grade 10 or below. Limit required qualifications to what is truly essential for the role. Negative Constraints: Do not use flowery metaphors or AI clichés like “delve,” “tapestry,” “passionate rockstar,” or “fast-paced environment.” Use direct, punchy, and professional language. |
Template 2: Update a JD for a Changed Role
Use this when an existing role has changed significantly and the current JD no longer reflects what the job actually is. Common triggers include new tools or technology, expanded or reduced scope, restructured reporting lines, role reclassification, or a previous JD that consistently attracted the wrong candidates.
| PROMPT — Update a JD for a Changed Role |
| Below is an existing job description that needs updating. Here is what has changed:Nature of change: [e.g., new tech stack, expanded scope, restructured team, reclassified seniority] What’s been added to the role: [INSERT] What’s been removed or reduced: [INSERT] New tools or systems the hire will use: [INSERT] Any changes to reporting lines or team structure: [INSERT]Using only the changes above, rewrite the job description with the following rules:Keep everything that hasn’t changed — do not rewrite sections that are still accurateUpdate only the sections affected by the changes listed aboveDo not invent new responsibilities or requirementsPreserve the original tone and brand voiceIf the job title no longer reflects the role, suggest a more accurate alternative[PASTE EXISTING JD HERE] |
Template 3: Build a Consistent JD for a Job Family
Use this when you need to create multiple JDs across levels (e.g., Analyst, Senior Analyst, Manager) and want them to feel like a cohesive set.
| PROMPT — Job Family Series |
| I need to create a series of job descriptions for the [INSERT JOB FAMILY] function at our company. There are [X] levels: [LIST LEVELS, e.g., Analyst, Senior Analyst, Lead, Manager]. For each level, write a job description that:Uses the same section structure across all levelsClearly differentiates scope, autonomy, and impact by levelAvoids repeating the same bullet points verbatim between levelsUses gender-neutral, bias-free languageHere is context about our company and team: [INSERT BRIEF COMPANY / TEAM CONTEXT]Generate all [X] JDs in sequence, clearly labeled by level. |
Template 4: Generate a Skills-Based JD
Use this when your organization is shifting to skills-based hiring and wants to move away from degree or years of experience requirements.
| PROMPT — Skills-Based JD |
| Write a skills-based job description for the role of [INSERT TITLE].Our organization is committed to skills-based hiring, which means:Do NOT list a specific degree requirement unless the role is legally regulatedDo NOT use “X years of experience” as a proxy for competencyDO list the specific skills, competencies, and outcomes we’re hiring forDO separate “Must Have” skills from “Nice to Have” skillsThe role involves: [BRIEF DESCRIPTION OF KEY RESPONSIBILITIES]Use plain language. |
Tips for Enterprise Teams
Beyond individual prompts, here are practices that help TA teams get consistent results at scale:
Standardize Your Prompt Variables
Before you start writing, create a short intake form for hiring managers to fill out. This is a simple document (a Google form or Word doc) with fields that map directly to your prompt template variables: job title, department, key responsibilities, must-have requirements, and team context. This reduces back-and-forth and keeps JDs accurate.
Build a Prompt Library
Save your best-performing prompts in a shared team document or wiki. Version-control them just like you would a JD template. Assign ownership to someone on the team to keep them updated.
Add Your Brand Voice to Every Prompt
Paste 2-3 sentences from your careers page or employer brand guidelines directly into your prompt. This helps the AI model match your company’s tone rather than producing generic copy.
Example addition to any prompt:
“Our company voice is [direct / warm / bold]. Here is an example of our tone: [PASTE SAMPLE COPY].”
Use AI Output as a First Draft, Not a Final Draft
AI tools are fast first-draft generators, not final editors. Always have a human reviewer check the output for accuracy, role-specific nuance, legal compliance, brand tone and voice, and alignment with current compensation bands.
Pair AI with a Bias-Detection Tool
AI tools can introduce subtle bias even when instructed to avoid it. Running your AI-generated JD through a dedicated bias-detection tool like Ongig’s Text Analyzer adds a reliable second layer of review.
What to Watch Out For
| Common AI JD Mistakes | How to Avoid It |
| Hallucinated qualifications | Always review requirements against the actual role. Remove anything not confirmed by the hiring manager |
| Generic “who we are” copy | Provide a real company description in your prompt rather than letting the AI invent one |
| Inflated requirements lists | Specify a maximum number of requirements in your prompt (e.g., “no more than 6 required qualifications”) |
| Inconsistent tone across JDs | Include a tone reference or brand voice example in every prompt |
| Subtle gender bias in language | Include explicit instructions to use gender-neutral language AND run output through a bias checker |
| Copy that looks AI-generated | Ask the model to “avoid AI clichés like ‘fast-paced environment’ and ‘wear many hats’” |
| JD is written for humans but invisible to algorithms. | Prompt the AI to incorporate industry-standard titles and skill keywords in top-performing job posts. |
Putting it into Practice
Many recruiters are now using AI to generate job descriptions. But the teams getting the best results are using AI combined with structured prompts, human review, and the right supporting tools.
Start with one or two of the templates above, adapt them to your company’s voice, and build your library from there. The upfront investment in good prompts pays off every time you need to fill a new role.
Always remember to check AI-generated JDs before posting them. To ensure unbiased job descriptions, use Ongig. Contact us to schedule a demo.
