Your business, out of your head.Founding clients: a few spotsSubscribe

Guides / Playbooks

Guide

Choosing AI tools for a small business: the boring checklist that works

The best predictor of AI tool regret is not picking a bad tool; it is picking any tool before your knowledge is portable. Evaluate on three axes (capability on your real work, lock-in risk, cost at the scale you will actually hit) and test with your own data before you pay.

The three axes.

  • Capability on your real work. Not the demo, your actual hardest case. A voice agent vendor demo handles "what are your hours"; the test is "does the MX wheelset fit a 2023 Sting" with a compliant answer.
  • Lock-in risk. What leaves with you if you cancel? If the answer is "nothing," the price is not the price. Your knowledge base, workflows, and history should live in files you own.
  • Cost at your real scale. Price the tier you will hit in month 6, not the teaser. In a real evaluation for a lending client (~35 vendors profiled, finalists scored on a weighted matrix), the flat-cheap option won the price criterion and still lost 4.10 to 3.35, on workflow fit, compliance, and lock-in.

The order of operations.

  1. Knowledge base first, tools second. A written brain makes every tool better and makes switching tools cheap. Tools first is how you end up rebuilding everything per platform.
  2. One general assistant before point tools. A $20-100/mo assistant wired to your files and email replaces several $300-400/mo single-purpose subscriptions for drafting, readouts, and research. Add point tools only for jobs it measurably cannot do.
  3. Test on your data, on a deadline. One week, your real messages and numbers, a pass/fail list written before the trial starts. If a vendor resists a real trial, that is the result.

Red flags that end the conversation.

  • Export costs extra or export does not exist.
  • "Trust the AI" instead of citations. No sources means you are the QA department, forever.
  • Per-seat pricing for a team of one. Owner-operator tools should price by usage, not headcount.
  • The demo cannot use your data. If it only shines on their sample data, it only works on their sample data.

Questions

Asked before reading this far.

What AI tools does a small business actually need?

Most owner-operators need less than they think: one capable general assistant (Claude or similar, $20-100 a month) wired to their email, files, and store data, on top of a written knowledge base. Point tools earn their subscription only when they do a job the general assistant measurably cannot.

How do I compare AI vendors?

Score every candidate on a weighted matrix across capability on your real work, lock-in risk, and cost at your true scale, then trial the finalists on your own data against a pass/fail list written in advance. This is the same method used in a real ~35-vendor evaluation for a lending client, where the finalists were scored on a weighted matrix.

How do I avoid AI tool lock-in?

Keep the knowledge outside the tools: your knowledge base, SOPs, and templates as plain files you own. Then any tool is replaceable, trials are cheap, and a vendor price hike is an inconvenience instead of a hostage negotiation.

Or skip the homework

This is what month one builds, done with you.

The knowledge base, the workflows, the tools wired, in four weeks, yours to keep. Priced by fit · Book a discovery call · See what changes in 30 seconds