ع
Start Topics Teams Reference What's new Saved
tracks

People & HR

Draft job posts, balance screen notes, rewrite policies, and build onboarding docs — consistent, inclusive, and fast — all from the Claude Desktop chat, no terminal required.

11 playbooks
playbooks

Full worked systems, not one-off prompts — each walks the whole workflow end to end. Pick one and follow it.

Build the competency framework every role inherits

Build one competency framework — named competencies, a level ladder, and the observable behaviors that show each level — that every job description, interview scorecard, performance review, and promotion case inherits, so the whole org judges people against the same bar instead of each manager's private idea of "senior".

medium ~1.5 hours to build, reused everywhere open the playbook

Write a job description that widens the pool

Turn a rough role idea into a clear, inclusive job description — genuine must-haves only, no gendered or exclusionary language, every quiet screen-out flagged for you.

easy ~20 min open the playbook

Draft a sensitive message you can stand behind

Draft a reorg note, hard feedback, or an org-change announcement carefully — honest, kind, no over-promising, names kept out, placeholders for what only you can confirm, and a shape you read word-for-word before it ships.

medium ~30 min open the playbook

Turn five interviewers into one fair verdict

Synthesize a scattered panel into one balanced read — shared strengths, shared concerns, the real disagreements kept visible, and 'culture fit' language flagged as bias, not evidence.

medium ~30 min open the playbook

Build a reusable onboarding-in-a-box

Build a role-specific first-week checklist and a 30-day plan — accounts, people to meet, docs to read, one clear early goal — grouped by day and reusable for every new hire in the role.

medium ~45 min open the playbook

Rewrite a policy into plain English

Turn a jargon-heavy policy into something a new hire understands on the first read — every rule kept, a one-line summary up top, and a human confirming nothing legal or factual shifted.

medium ~45 min open the playbook

Run a fair, calibrated performance-review cycle

Anchor every review to one competency bar, weight evidence over eloquence, keep real disagreements visible, run a deliberate bias pass, and calibrate ratings across people — so the same rating means the same thing and the manager owns the number.

advanced ~half a day per cycle open the playbook

Turn an engagement survey into an action plan

Cluster a CSV of open-text survey comments into the handful of themes that matter — weighted by how many people raised them, quoted anonymously, and turned into owned, dated actions — while protecting the anonymity that made people answer honestly in the first place.

medium ~1 hour open the playbook

Run a respectful exit and keep what you learn

Build the offboarding box once: a dignified exit checklist, a least-access removal list sequenced so security-critical access closes first, a real knowledge-transfer capture before the last day, and a fairly synthesized exit interview — reused identically for every leaver.

medium ~45 min open the playbook

Run a hire from open to offer

Run a whole role end to end — inclusive JD, structured interview kit and scorecard, a fairly synthesized panel, a bias check, and the offer and rejection comms — one consistent, defensible system where the human always makes the call.

advanced ~half a day open the playbook

Run an org change without losing the room

Roll out a reorg, a policy change, a leadership change, or a return-to-office mandate like a launch — one source-of-truth narrative, a manager enablement pack, a sequenced leaders → managers → all-hands rollout, an honest anticipated-questions FAQ, and a real feedback loop — so the change lands with the trust intact and the human owns every decision.

advanced ~half a day open the playbook
Go deeper

People & HR with Claude, in depth

The playbooks above are single jobs. These are the long reads behind them — how the work fits together across a quarter, what good looks like, and the judgement calls the prompts assume you have already made. Read in any order.

Guardrails

The non-negotiables before any of this ships. Read the operating guide for the full data, brand, legal, and rollout playbook.

  • People data stays in an approved workspace. Employee and candidate details — names, notes, comp, reviews, exit reasons — go only into a tool your company has cleared for people data, and you approve each read in the 'Ask permissions' prompt. Build templates with [placeholders]; apply them to real people only inside your approved system.
  • Claude drafts; humans decide. Every hire, promotion, rating, and exit is a person's call. Claude balances the inputs, surfaces bias, and drafts the comms — it never makes the people decision, and it never owns the consequence.
  • Hunt for bias, then calibrate. Ask Claude to flag 'culture fit', personality, and anything not tied to the actual job — then weight evidence equally and calibrate across people, so a verdict rests on the work, not on who wrote the most or shipped last.
  • Verify the load-bearing facts, and route the regulated ones. Claude states comp numbers, legal entitlements, and policy rules with total confidence whether or not they're right. Verify every one, and send anything touching employment law, protected characteristics, or comp past a real person before it ships.
  • One sign-off gate, named owners. At a team's scale, name who approves: HR owns fairness and process, legal owns anything regulated, the manager owns the decision. The engine fills the queue; it never skips the reviewer.
  • Adapt for Arabic, don't translate. Bilingual HR — JDs, policies, sensitive messages, exit conversations — is authored natively in Arabic with the same bias scan and the same sign-off, not a machine-translated copy of the English.
Read the operating guide

Questions people ask

Is it safe to put employee or candidate data into Claude?
Never paste private employee or candidate details into a tool your company hasn't approved — confidentiality comes first. Work from the notes inside a workspace your company allows for people data, let Claude read the local copy through the Desktop file pane (approving each read in the 'Ask permissions' prompt), and keep names out where you can.
Does this replace my recruiters or make the hiring decision?
No. It drafts, summarizes, and flags possible bias, but a human always makes the final hire, promotion, or exit call. Claude balances the inputs; the people in the room decide.
Can it help with legal or policy wording?
It can rewrite policies in plain English and draft sensitive messages, but send anything legal to a real lawyer before it goes out. Have a person confirm nothing factual or legal shifted in a rewrite.
What's a realistic first task?
Balancing scattered interview notes into one fair summary: hand Claude five interviewers' notes and ask for the shared strengths, shared concerns, and where they disagree — weighted equally, not by who wrote the most.