November 19, 2026 · AI writer · Oak City Intelligence

A One-Page AI Policy for a Small Shop

A flat vector illustration in Oak City Intelligence's house style for "A One-Page AI Policy for a Small Shop."

Why Write One at All

Odds are somebody on your staff already pastes customer emails into ChatGPT to get a reply drafted faster, and usually that helps. Without a rule, though, nobody knows which account they used, what went into it, or whether anyone read the reply before it went out. A policy settles those questions once, so you aren't settling them one mistake at a time.

Corporate AI policies often run long and assume an IT department. A shop of three to fifteen people needs one page that a new hire can read on their first morning. Copy the one below, change the bracketed parts, and print it. Done.

The Policy

How we use AI at [Business Name]

1. Which tools. We use [ChatGPT / Claude / Gemini / Copilot] on the business account only: the one signed up under our business email, or our business plan. Don't use a personal account for work, and don't sign up for new AI tools without asking [owner's name] first.

2. What never goes in. Card numbers. Social Security numbers. Passwords or login codes. [Patient health details.] [Anything a client tells us in confidence about their legal matter.] When in doubt, leave the name out and describe the situation instead.

3. A person reads it first. Nothing an AI writes goes to a customer, a supplier or the public until you have read every word. You are the one sending it, and you answer for it.

4. Check the facts it can't know. Prices, dates, hours, warranty terms, and anything about a specific job get checked against our own records before they go out. AI tools make these up when they don't know them, and they sound sure either way.

5. Ask when it matters. Refunds, complaints, legal questions and anything that commits us to money go to [owner's name], with or without AI.

6. Say so if asked. If a customer asks whether a message was written with AI help, tell them the truth.

Reviewed [date]. We look at this again every six months.

Introducing It

Hand it out at the next staff meeting and keep the discussion to ten minutes. Ask who already uses an assistant at work and what for, and listen without scolding, because the answers tell you which rules your people will actually bump into. Then walk through rules two and three with a real example from your own week: a customer email, the draft an assistant would write, and the one wrong detail someone would need to catch.

Set up the business account before the meeting, not after. A rule that says "business account only" with no business account to use just pushes people to keep using their personal ones quietly.

Keeping It Alive

Put a reminder on your calendar for six months out, on a morning you know will be quiet, and title it "Review AI policy" so the file is easy to find when it comes up. Tools change, your staff changes, and the line about which assistant you use will go stale first. When something goes wrong in between, a reply sent with a wrong price or a customer upset to learn a robot drafted their email, add a line that would have prevented it. A policy that grows from your own mistakes stays short and gets read, while one copied once and never changed gets signed and forgotten.


Written by Oak City Intelligence's AI writer. Reviewed before publication.