AI for HR
HR is high-stakes, high-sensitivity work where AI can either be a force multiplier or a liability. The difference is using it for scaffolding (drafts, structure, templates) and keeping the judgment, empathy, and final word with you. AI helps you write more carefully — not faster, but with fewer blind spots.
~/.claude/skills/What AI handles well
Performance review feedback that's specific (not generic)
The problem. Annual reviews are due. You have 40 to write. The default is vague compliments and "areas for development" that don't mean anything.
What AI does. Feed AI the manager's notes + employee's self-review. Have it draft feedback that's grounded in specific examples, ties to actual behaviors, and surfaces the contradiction between manager and self-perception.
Internal comms (announcements, policy changes, restructures)
The problem. Leadership decided to change the WFH policy. You have 2 days to announce it. The first draft from leadership is corporate-speak. People will read between the lines.
What AI does. Use the announcement prompt. AI drafts something that names what's changing in line 1, explains why honestly, what people need to do, and where to ask questions.
Onboarding plans that don't feel templated
The problem. New hire starts Monday. Your onboarding template is 50 generic steps. They'll feel like a number on day 1.
What AI does. Use AI to tailor a 30-60-90 plan based on the role, team, and what the new hire said in their interview. Output: a real plan referencing real people, projects, and goals.
Hard conversation prep (PIPs, terminations, comp discussions)
The problem. You have a hard conversation Monday. You need to walk in clear, kind, and prepared — not winging it.
What AI does. Use the tough-conversation prompt to sharpen what you actually need to say (no euphemisms), the specific examples, what to avoid, and how to handle defensive reactions.
Policy drafts that fit your culture
The problem. You need a policy on AI use, remote work, expense limits, whatever. Generic templates from HR consultants don't fit your culture.
What AI does. Have AI draft starting from your existing values doc + a few example scenarios you want the policy to handle. Iterate until it sounds like your company, not generic HR-speak.
Employee survey design + analysis
The problem. You're running a culture survey. Generic 1-10 scale questions produce vanity stats. Open-text gets ignored. You want signal.
What AI does. Use the survey-design prompt. Limit to questions that'll inform a decision. After collection, use AI to thematically analyze open-text responses and surface dissent.
Manager training content
The problem. Your managers want training. Off-the-shelf programs feel generic. Building from scratch takes months.
What AI does. Use AI to draft scenarios specific to your business (your common conflict patterns, your typical performance issues). The training feels relevant because it is.
Job descriptions that don't sound like every other JD
The problem. You're backing up a hiring manager. Their JD is a wall of "rockstar, fast-paced, wear many hats."
What AI does. Use the JD prompt. Real work the person will do, honest tradeoffs, what success looks like, real comp range. Candidates who fit will recognize themselves.
Your AI stack
Start with the foundation. Add specialized tools as the work calls for them.
Foundation LLM
Specialized add-ons
Prompts ready to use
Get started in 35 minutes
Pick an LLM and load your handbook + culture docs
15 minClaude Project or Custom GPT with: your culture deck, your benefits, your management philosophy, common scenarios. Any sensitive HR prompt now starts pre-loaded.
Run the announcement prompt on your next internal change
10 minWhatever change is coming next — a policy update, restructure, etc. Draft with AI. Notice how much sharper it is than the standard "we're excited to announce..." opener.
Use the tough-conversation prompt before your next 1:1 with a struggling employee
10 minDon't skip the prep. AI helps you walk in with the actual sentence you need to say, not a softened version that protects you instead of helping them.
Common mistakes
AI-generated performance feedback without re-personalizing. The employee can tell. "Demonstrates strong communication skills" written by AI lands as zero. Specificity is the whole point.
Using AI to summarize sensitive conversations. The compression loses the nuance — the moment of frustration, the half-said thing — that's often where the actual signal lives.
Pasting confidential employee data (names, performance issues, comp, mental health context) into public AI. Use approved enterprise tiers or anonymize.
Generating HR policies that don't fit your culture. "Reasonable expense limits" means different things at different companies. Generic policies signal you didn't actually think about it.
AI-screening that quietly perpetuates bias. AI scores against patterns. If your historical hiring data is biased, AI will replicate the bias. Audit what it's selecting for, especially in screening tools.