Practical Recruiting Skills
How to Use Claude for Recruiters: Sourcing, Screens, and Outreach
Better drafts, sharper questions, and a clear line between writing help and hiring decisions.
Anthropic's Claude can turn a hiring manager's rough notes into a job description, generate a consistent set of screen questions, draft candidate outreach, and organize your pipeline notes after a call. This page covers how to use Claude for recruiters across those four workflows, with worked prompts you can copy, a starter path for learning the underlying skill, and a clear answer to the question every recruiting team eventually asks: where does the writing assistant stop and the human decision-maker start.
How to use Claude for recruiters comes down to four repeatable workflows: drafting job descriptions from a hiring manager's notes, generating structured screen questions tied to named competencies, writing candidate outreach and follow-up messages, and organizing pipeline notes into a consistent format after a call. Claude works as a drafting and research assistant, not an evaluator. It has no access to your applicant tracking system, no knowledge of your actual candidates, and no ability to judge fit. Every evaluation decision, from who advances to who gets an offer, stays with the recruiter.
Published · Last reviewed
Free lesson · no account neededMeet ClaudeWhy Claude?6 minStart the free lessonKey Takeaways
- Claude drafts job descriptions, screen questions, outreach messages, and pipeline note summaries; it does not evaluate or rank candidates.
- A drafted job description or screen guide still needs human review for accuracy, fairness, and fit with the actual role.
- Using Claude to write is a different category from using AI to score or filter applicants, which carries separate legal considerations.
- The same prompt template, reused across a recruiting team, makes pipeline notes and screen guides consistent across different recruiters.
- Specific prompts, naming the role, the competencies, and the tone, produce usable first drafts; vague prompts produce generic ones.
- The starter path below teaches the prompting habits that make every one of these workflows faster and more reliable.
The Recruiting Workflow, With and Without Claude
| Task | Without Claude | With Claude |
|---|---|---|
| Job descriptions | Written from scratch or copied from an old posting and lightly edited | Drafted from the hiring manager's rough notes, then reviewed and edited by the recruiter |
| Screen questions | Improvised per interviewer, so the questions vary from candidate to candidate | A structured set tied to named competencies, reused consistently across the team |
| Candidate outreach | A generic template or a blank cursor for every new candidate | A first draft built from the role and the candidate's background, then personalized before sending |
| Pipeline notes | Scattered shorthand that looks different depending on who wrote it | Organized into the same structure every time, easier to compare across candidates |
Copy and Paste
Prompts You Can Use Today
Draft a job description from rough notes
Here are the hiring manager's rough notes for a [job title] role: [paste notes]. Draft a job description with a summary, responsibilities, required skills, and preferred skills, written in plain language for a candidate audience. Flag any requirement that seems unnecessarily restrictive given the notes.
Swap in the actual notes and title; always review the flagged requirements with the hiring manager before posting.
Generate a structured screen guide
Generate eight behavioral and situational interview questions for a [job title] role. Focus on these competencies: [competency one], [competency two], and [competency three]. For each question, include a short note on what a strong answer sounds like so different interviewers can calibrate consistently.
Reuse the same competency list across every candidate for this role to keep the screen fair.
Write a candidate outreach message
Write a LinkedIn outreach message to a [role, for example senior product designer] candidate. The opportunity is at a [company stage or industry, for example Series B fintech] company. Mention what makes this specific role interesting, not generic praise. Tone: direct and respectful, not salesy. Keep it under 150 words.
Add one real detail from the candidate's background before sending; a fully generic message reads as one.
Summarize a phone screen into pipeline notes
Here are my raw notes from a phone screen with a [job title] candidate: [paste notes]. Organize them into three sections: strengths observed, areas to probe further in the next round, and open questions for the hiring manager. Keep it factual and specific to what was said, not general impressions.
Use this same three-section template for every candidate in the pipeline so notes are easy to compare later.
Write a respectful rejection email
Write a rejection email for a candidate who reached the [stage, for example final panel] stage for a [job title] role. The message should be clear about the outcome in the first two sentences, avoid vague language about 'moving forward with other candidates,' and leave a positive impression of the company.
Personalize with one specific detail about the interview if you can; templated language is easy to spot at this stage.
Check a draft job description for bias language
Review this draft job description for language patterns associated with demographic bias, such as gendered adjectives or requirements that exceed what the role actually needs: [paste draft]. List each flagged phrase, explain why it might narrow the candidate pool, and suggest a neutral alternative.
Treat the flags as a prompt to review, not a compliance sign-off; a second human read still matters.
Start Here, Free
The Starter Path for Recruiters
Three free lessons that teach the prompting habits behind every workflow on this page, from job descriptions to pipeline notes.
Core Workflows
What Recruiters Actually Use Claude For
These are the writing-heavy tasks where Claude saves real time without touching the judgment calls that belong to you.
Job Description Drafting
Paste a hiring manager's rough notes and ask Claude to turn them into a structured job description. You can request an inclusive-language pass and ask it to flag requirements that read as unnecessarily restrictive.
Structured Screen Questions
Claude generates a consistent set of behavioral and situational screen questions tied to the competencies you specify. Consistent questions reduce variance in how different candidates get assessed.
Candidate Outreach
Claude drafts sourcing messages and follow-up emails. Give it the role, the candidate's background, and the tone you want, and it produces a first draft for you to personalize and send.
Pipeline Notes and Summaries
After a screen call, dictate or paste your raw notes and ask Claude to organize them into a structured summary. This keeps records consistent across the team and easier to review later.
Role Research and Market Context
Ask Claude to explain what a role typically involves, which skills are standard versus rare, or how a title maps across industries. Useful when you're recruiting outside your usual area.
Bias Awareness Check
Claude can review a draft job description or a set of screen questions and flag language patterns associated with demographic bias. Treat this as a prompt to review, not a guarantee of compliance.
Where the Lines Are
Bias, Fairness, and Where Judgment Stays Human
Claude is a drafting and research tool. It does not evaluate candidates. It has no access to your applicant tracking system, no knowledge of your actual candidate pool, and no ability to assess whether a specific person is right for a specific role. Every evaluation judgment, from who advances past the screen to who gets an offer, belongs to the recruiter and the hiring team.
Bias risk in AI-assisted recruiting is real and worth taking seriously. Claude can help you write more consistent language and can flag phrases associated with demographic bias, but it reflects patterns in its training data, which includes decades of hiring norms that carry their own biases. A Claude-drafted job description is not automatically neutral; it is a starting point that still needs a careful human read.
The practical way to handle this is to treat every bias flag as a question, not an answer. If Claude flags a requirement as restrictive, decide with the hiring manager whether it is actually necessary for the role. If it suggests neutral phrasing, check that the meaning didn't shift. The tool surfaces things to look at; it doesn't certify that a document is fair.
The same caution applies to screen questions. A consistent question set reduces one kind of variance, interviewer improvisation, but it doesn't remove the need for interviewers to score answers fairly. Structure helps; it isn't a substitute for training your team on how to run a fair interview.
Compliance
Legal Risk: Drafting Is Not Evaluating
In many jurisdictions, using automated tools in hiring decisions triggers specific compliance requirements, and the rules differ by location and by what the tool is actually doing. This is why the distinction between drafting and evaluating matters so much: using Claude to draft a job description or outreach copy is a different category from using AI to score, rank, or filter candidates.
Drafting a job description with Claude's help is low-risk. It's a writing task; the recruiter reads, edits, and approves the final version before it goes anywhere near a candidate. Using an AI system to automatically screen resumes, score interview answers, or filter applicants out of a pipeline is a different kind of decision, and one that fair-employment law in various places treats differently.
If your team is considering AI-assisted screening or scoring beyond drafting, that's a conversation for legal counsel, not something to decide based on what a writing assistant happens to be capable of. Keep Claude in the drafting and research lane, and the legal exposure stays limited to the same standards you'd apply to any written material a human produced.
Getting Started
The Starter Path for Learning This
The workflows above all rest on the same underlying skill: writing prompts that give Claude enough context to produce something worth editing instead of something generic. Three free lessons build that skill in order, using recruiting-relevant examples along the way.
Start with Meet Claude, from the Why Claude? course. It's a plain introduction to how Claude works and where it's strong or limited, useful before you start feeding it hiring manager notes or candidate details.
Next, Make Claude Sound Like You, also from Why Claude?, covers how to get Claude's drafts, whether it's an outreach message or a rejection email, to sound like something you'd actually send rather than something obviously templated.
Then Context Is Everything, from Strategic Prompting, is the lesson that matters most for recruiting work specifically: it teaches you how to hand Claude the details, role, competencies, candidate background, tone, that turn a vague job description draft into one worth sending to a hiring manager for review.
Better Prompts, Better Drafts
Giving Claude the Right Context for Recruiting Work
Specificity matters more in recruiting prompts than in most other use cases. "Write a job description for an engineer" produces generic output you'll rewrite from scratch. "Write a job description for a backend engineer who will own our payments infrastructure, working on a small team, with strong Rust experience required and Go preferred" produces something worth editing instead of replacing.
The same logic applies to screen questions and outreach. Name the actual competencies you're assessing, not just the job title. Name what makes the specific role interesting to a candidate, not generic praise about the company. The more real detail you give Claude, the less generic the draft comes back.
Templates compound the value over time. Build a small library of prompts for the tasks you run repeatedly, screen guides, outreach messages, rejection emails, pipeline note summaries, and share them across your recruiting team. When everyone uses the same underlying structure, the output stays consistent even when different people are writing it, which matters as much for candidate notes as for candidate outreach.
Always edit before sending. Claude's outreach and rejection drafts often need a personal detail, a specific project reference, or a tone adjustment to feel human rather than templated. The draft gets you past the blank page; the edit is what makes it yours.
Keep This in Mind
Limits and Professional Responsibility
A few boundaries worth setting before you build Claude into your recruiting workflow.
Confidential Candidate Data
Know your organization's data handling policy before pasting resumes, candidate contact details, or compensation history into any AI tool. Some information shouldn't leave your applicant tracking system at all.
Never the Decision-Maker
Don't use Claude to score, rank, or filter candidates. Evaluation judgment stays with the recruiter and hiring team, every time, regardless of how good a drafting assistant it is for job descriptions and questions.
Check Before It Reaches a Candidate
Read every outreach message, rejection email, and job description before it's sent. A draft that's factually wrong about the role or the company is worse than no draft at all.
One Tool Among Several
Claude is a writing and research assistant inside a larger recruiting process that includes your applicant tracking system, structured interview scorecards, and your organization's hiring policies. It doesn't replace any of those.
Learn to Use Claude Effectively as a Recruiting Tool
Reading about these workflows and running them reliably under real recruiting pressure are two different things. The Nightschool AI curriculum is hands-on from the first lesson, building the prompting habits that make job descriptions, screen guides, and outreach hold up once you're actually using them.
Frequently Asked Questions
Can Claude screen candidates for me?
No. Claude is a drafting and research tool. It cannot access your candidate pool, review resumes against a role, or make any evaluation judgment. Using AI to score or filter candidates raises legal and ethical considerations well beyond what a writing assistant is designed for. Claude helps you write and organize; humans evaluate and decide.
Will Claude make my job descriptions less biased?
Claude can flag language patterns associated with demographic bias, such as certain adjectives or unnecessarily narrow requirements. That's useful as a prompt to review your draft more carefully, but it's not a guarantee. Claude reflects its training data, which contains historical hiring norms and their biases. Treat its suggestions as one input, not a compliance check.
How do I get Claude to write good outreach messages?
Give it three things: what makes the role genuinely interesting, a summary of the candidate's background based on what you know, and a clear tone instruction. The more specific you are about what differentiates this role from a generic one, the more specific the message comes back. Always review and add a personal detail before sending.
Can Claude help me write structured interview questions?
Yes, this is one of the higher-value uses. Describe the role and list the competencies you want to assess, then ask for behavioral or situational questions tied to each one. You can also ask Claude to describe what a strong versus weak answer looks like, which helps interviewers calibrate consistently across candidates.
Is using Claude in hiring legally risky?
Using Claude to draft a job description or write outreach copy is low-risk; it's a writing task the recruiter reviews before it reaches anyone. Using AI to evaluate, score, or filter candidates is a different category and may trigger requirements under fair-employment law in various jurisdictions. If you're considering AI-assisted screening, that's a question for legal counsel.
How do I keep pipeline notes consistent across my team?
Build a shared prompt template that every recruiter uses after a screen call, specifying sections like strengths observed, areas to probe further, and open questions for the next round. When everyone runs their raw notes through the same structure, the resulting records become much easier to compare across candidates and across recruiters.
Can Claude write rejection emails that feel respectful?
Yes. Give it the role, a note on what stage the candidate reached, and the tone you want. Ask for a message that's clear about the outcome early, specific rather than vague, and leaves a positive impression. Claude tends to produce serviceable drafts that need light editing and a personal detail to feel less templated.
What should I not use Claude for in recruiting?
Don't use it to evaluate candidate quality, rank applicants, or make any selection decision. Don't feed it confidential candidate data without understanding your organization's data handling policy. Don't treat its bias flagging as a legal audit. Claude is a writing and research assistant; keep it in that lane and the workflows above are straightforward.
Related reading
Put These Skills to Work
Hands-on practice is how the prompts stop feeling awkward and start feeling fast.