“We need an AI strategy” is a sentence that produces meetings. “Our quoting process takes four hours and we do it eleven times a month” is a sentence that produces results.

If you run a small business or a nonprofit, you do not have spare capacity for a transformation initiative. You have a Tuesday. So here is a method for finding where AI will actually pay off in your organization, which takes about an hour and requires no technology at all.

Step one: list what you actually repeat

Open a blank page and write down every recurring process in your organization. Not projects — processes. Things that happen again and again: writing proposals, answering the same twelve customer questions, producing the monthly report, onboarding a new hire, drafting the newsletter, researching grants, chasing invoices, updating the website.

Aim for fifteen to twenty. Most people are surprised by the length of this list, which is itself informative.

Step two: score each one on three things

For each process, note three numbers:

  • Frequency — how many times a month does this happen?
  • Time — how long does one round take, honestly, including the part where you avoid starting it?
  • Tolerance — if the first draft were 80% right and needed editing, would that be fine? Or does this need to be perfect and defensible on the first pass?

Multiply frequency by time to get monthly hours. Then look at tolerance, because it is the deciding factor people skip. A task that eats twelve hours a month but must be flawless and legally precise is a bad first candidate. A task that eats six hours a month where a solid draft saves you most of the work is an excellent one.

Step three: pick three and ignore everything else

Sort by monthly hours, filter for high tolerance, take the top three. Those are your projects. Write them on something you will see.

The discipline here is the ignoring. Every organization I have watched try this has been tempted to start with the most interesting opportunity rather than the most repetitive one. The interesting one makes a better story at a conference. The repetitive one gives you back your Thursday afternoons, which is what you actually wanted.

Step four: write down how you do it now

Before you bring AI anywhere near the process, document the current version. Plain steps, in order, the way you would explain it to a new hire.

Two things happen when you do this. First, you find dead steps — approvals nobody reads, fields nobody uses, a spreadsheet that exists because of a problem you solved years ago. Removing those is free improvement that requires no technology. Second, you now have the exact context an AI tool needs to be useful. “Here is how we write proposals, here are two good examples, draft one for this client” produces something usable. “Write me a proposal” produces something generic.

What good looks like after a month

Not “we are an AI-forward organization.” Something more like: the quote that took four hours takes ninety minutes, the newsletter gets written on the day it was scheduled instead of four days late, and the twelve repeated customer questions have solid answers that anyone on the team can send.

Three fixed processes beat one grand strategy, every time.

Then — and only then — pick the next three.