The average AI-adopting small business now runs about five AI tools. Some earn their subscription every day. Some are shelf-ware with a monthly charge. Here’s the rubric we use with clients — three questions per tool, no vendor pitch survives all three.
AI tools for small business, question one: does it sit inside a process?
A tool you have to visit is a hobby. A tool that lives inside something you already do daily — your inbox, your phone line, your invoicing — is infrastructure. When we audit a client’s stack, the first cut is simple: if using the tool requires remembering it exists, it goes on the cancel list.
Question two: buy it or build it?
Buy when your need is generic — writing assistance, meeting notes, image cleanup. The market is competitive and cheap. Build (or have it built) when the value comes from your data and your workflow: matching your payments to your invoices, answering your phone with your prices, triaging your inbox. Generic tools can’t do those, because the hard part isn’t the AI — it’s the wiring. Pricing for custom work has fallen dramatically; a scoped build that once required a development team now lands in the range a small business can actually budget.
Question three: what happens when it’s wrong?
Every AI is sometimes wrong. The question is whether the failure is visible and recoverable. A drafted email you review before sending: safe. An automation that silently marks invoices paid: dangerous without monitoring. Skip any tool whose failure mode is invisible.
Run your current stack through those three questions and you’ll usually find one subscription to cancel, two to keep, and one process — usually the phone or the admin pile — that deserves a real build.
A consult costs thirty minutes and nothing else. Tell us what’s eating your week.
Book a consult