Field Notes · 06
What a small business should hand to AI, and what it should keep
A working checklist for owners: five kinds of work a machine can take today, four it should stay away from, and five questions to ask before customer data leaves the building.
Most advice about AI for small business falls into two piles. One says it will change everything by Friday. The other says it is a fad.
The owners we talk to are tired of both. They want an answer to a plain question: what should I do with this?
Here is the list we work from. It comes from running the back office of a working pediatric practice and from fifteen years of cleaning up after software that promised too much.
Hand these over
Retyping. A person reads information in one place and keys it into another. A machine can do that first pass. Emailed work orders go into a job sheet. Web form submissions go into a customer record. Receipts go into a ledger. A person reviews the exceptions and spot-checks the rest.
First drafts of routine messages. Appointment reminders, quote follow-ups, review requests, and answers to the nine questions everyone asks. The machine drafts in your voice from examples you supply. A person sends.
Sorting the inbox. The machine reads each incoming message and labels it: new lead, existing customer, vendor, invoice, junk. That alone gives some owners an hour a day back.
Summaries. A long email thread, a forty-page contract, or a month of customer comments becomes one page. The source passages are marked, so you can check them.
The weekly report nobody has time to build. Every Monday morning, the numbers from two or three systems land on one page.
Look at the pattern. Each job repeats. The inputs are text. A mistake is cheap to catch, and a person sees the result before it matters.
Keep these
Anything where the relationship is the product. The call after a job goes wrong. The conversation with a family in a hard season. Customers forgive a slow reply from a person. They do not forgive a fast reply from a machine that misses the point.
Final say on money. Let a machine prepare the invoice, flag the late account and draft the estimate. A person approves anything that charges, refunds or commits you.
Professional judgment. Diagnosis, legal advice, structural decisions and tax positions stay with people. A model can help a licensed person work faster. It cannot hold the license or carry the liability.
Facts you cannot check. These systems sometimes state false things with full confidence. If nobody on your team could tell a wrong answer from a right one, the task is a poor fit.
Before any customer data goes anywhere
Ask these five questions of any tool, and of any person who proposes one. That includes us.
- Which company's servers will see this information?
- Is it used to train their models? Where is that written down?
- How long is it kept, and can it be deleted on request?
- If the data is regulated, such as health, financial or student records, will the vendor sign the agreement your regulator expects?
- What happens to the workflow if this vendor doubles its price or shuts down?
A vendor who answers all five quickly and in writing is worth a conversation. Hesitation on the second question is a reason to walk away.
How to begin
Pick one job from the first list. Choose the one your staff complains about most, because they will help you get it right.
Run the machine beside the person for a few weeks and compare the results. Expand only when the comparison bores you.
That pace will feel slow next to the headlines. It is the pace that builds things that still work in five years, and that is the only kind worth paying for.
Have a question about your own site? Send a note and we will take a look.
Ask us about your siteGatorGeeks · West Palm Beach, Florida
gus@gatorgeeks.com · gatorgeeks.com/contact