Where AI helps an owner who hires a few times a year
You're not an HR department. You're hiring a tech, a crew lead, a line cook or a front-desk person, usually in a hurry, between jobs. The slow parts are writing a post that attracts the right people, screening consistently and remembering what you asked each candidate. AI is good at all three. It's not good at judging people, and in some places using automated tools to screen candidates comes with legal requirements.
Step 1: the job post prompt
Write a job post for a [role] at a [type of business] in [city or area]. Facts: pay range [$X to $Y], hours [details], must-haves [license, driving record, lifting], nice-to-haves [list], the hardest part of the job is [honest answer], and after 90 days a great hire will [outcome]. Plain language, under 300 words, no buzzwords like "rockstar" or "fast-paced family." Include how to apply. Don't add requirements I didn't list.
Then check it. Remove anything that could put off qualified people for reasons unrelated to the job, such as age-coded wording ("young, energetic team") or requirements you don't truly need. Make sure pay and other details meet the job-posting rules where you hire. Several states and cities require pay ranges in job ads.
Step 2: a phone-screen scorecard
From this job post: [paste], create a 15-minute phone-screen script with 6 questions that test the must-haves and the hardest part of the job, and a scorecard rating each answer 1 to 3 with a description of what a 1, 2 and 3 answer sounds like. Questions must be job-related only.
Use the same script for every candidate and note scores as you go. Consistency is fairer to candidates, and it makes your decisions easier to explain.
Step 3: interview questions and a work test
- Ask AI for behavioral questions tied to the hard parts of the job: "Tell me about a time a customer was upset about the price."
- Ask for a short, paid or realistic work test where it fits: a mock service call, a sample estimate, a prep task.
- After the interview, paste your own notes (not recordings, unless the candidate agreed) and ask AI to organize them against the scorecard. You still make the call.
The legal line: AI helps, a person decides
- Don't let AI rank, reject or score candidates automatically unless you've checked the rules where you hire. In New York City, Local Law 144 requires a bias audit and candidate notices before employers use automated employment decision tools (NYC DCWP). Other states and cities have their own rules, and more are coming.
- Anti-discrimination law applies however a decision is made. Using a tool doesn't move responsibility off you.
- Never ask, or let AI ask, about age, family plans, religion, health, disability, national origin or other protected traits.
- Keep applicant data private. Remove names and contact details before pasting résumés into general AI tools, or use a business plan whose data terms you've checked.
- When in doubt, ask your employment attorney or HR advisor. The cost of a quick question is small compared to a claim.
After the hire: onboarding
The same tools make onboarding faster: a first-week checklist, SOPs written from recordings (AI for SOPs), and a simple training plan (train employees to use AI).
Hiring, onboarding and training often take more owner time than anything else in a growing business. Your business's AI plan shows where AI can take hours off each step for your kind of business, and where to keep a person firmly in charge.