AI at work · For owner-led employers
Where AI can safely take on routine HR
AI is good at reading, drafting and organizing. It is not accountable for a decision about a person. Here is where that line sits for a small employer, and what to put in writing before your team starts using it.
The short version
Use AI to get to a first draft faster: job descriptions, policy language, interview guides, onboarding checklists, summaries of long regulations. Keep a person responsible for anything that changes someone's job, pay or record: hiring, discipline, termination, accommodation and leave decisions. And never put confidential employee information into a tool your company hasn't approved.
Where AI helps, with a person reviewing
| Task | What AI does well | What a person must check |
|---|---|---|
| Job descriptions | Turns a messy list of duties into a clean, consistent format. | Essential functions, physical requirements, exempt or nonexempt status, and pay range disclosure where your state requires it. |
| Policy drafts | First drafts of handbook sections in plain English. | That the policy matches each state you employ in and how you actually run. Generic drafts often miss state leave and final-pay rules. |
| Research | Summarizing statutes, agency guidance and deadlines. | The primary source. Models can state an outdated rate or a rule from the wrong state with full confidence. |
| Interview guides | Structured, job-related questions and scoring rubrics. | That no question touches a protected characteristic. Check a question. |
| Onboarding | Checklists, welcome emails, first-week schedules. | Deadlines: I-9 timing, new-hire reporting and state notices. |
| Employee questions | Answering "where do I find the PTO policy?" from your own documents. | That it answers only from approved documents and hands sensitive questions to a person. |
Where a person decides
- Hiring and screening. Automated resume screening can screen out protected groups without anyone intending it, and the employer is responsible for the outcome. Some jurisdictions now regulate these tools directly: New York City requires bias audits and notice for automated hiring tools, and Illinois law, effective January 1, 2026, covers AI used in employment decisions.
- Discipline and termination. AI can help write factual documentation. It shouldn't decide who is disciplined or let go, and its draft must be checked against what actually happened.
- Accommodation and leave. These turn on an individual conversation and on medical information that belongs in a separate, confidential file.
- Pay decisions. Raises, classification and pay equity need a person who can explain the reasons later.
Five policies to have in writing first
- Approved tools. Which AI tools staff may use for work, and which accounts.
- Data rules. No employee medical, payroll, Social Security or performance details in unapproved tools. Customer data follows the same rule.
- Human review. Anything AI drafts that goes to an employee, customer or agency is reviewed by a person who owns it.
- Disclosure. When you tell applicants or employees that AI is involved, and when the law requires it.
- Records. Keep the final, human-approved version, not just the AI draft.
How we use it
We use AI to research federal and state rules quickly and to draft. Every finding is checked against the primary source, and a certified HR professional reviews and signs off on everything we send. That's the same standard we recommend to clients. The Employer Checkup includes a short map of where AI can safely take routine HR work off your plate.
Practical HR guidance, not legal advice. AI rules for employment are changing quickly; check your state and city before deploying a tool that screens or evaluates people.