My wife and I had a problem at dinner. I had something working before bed.

Every weekend the same hour disappeared. Deciding what we were eating that week, checking what was already in the house, writing the list, then getting to Tuesday and realizing the thing we planned needed two hours we did not have.

It is a small problem. It is also one that came back every seven days forever, which is the kind worth fixing.

So we built something for it. It asks what we are in the mood for and what is already in the fridge, and then it emails the week: the menu, the grocery list in the order the store is laid out, and what to cook ahead on Sunday so the weeknights are short. The meals land on the calendar with the steps attached, so at five o’clock nobody has to remember anything or make a decision.

The hour is now about fifteen minutes.

The part I did not expect was the second thing it gave back. It was never really the hour. It was that dinner stopped being an open question hanging over the day. The decision got made once, on Sunday, by something that does not get tired at four in the afternoon and suggest going out.

That is the test I use now before building anything, and it is the one I would tell you to use in your business. Not “what is the most impressive thing I could automate.” It is: what do I do every single week that I already know the shape of, that I would never miss if it happened without me?

The answers are almost always boring. The list that gets made every Monday. The same four questions you ask every new customer. The reminder you send before an appointment. The invoice that goes out after a job closes. Nobody puts those in a plan because they are too small to feel like a project, and together they are most of the week.

Start with the one that annoyed you most recently. Not the biggest one. The one you thought about this week while doing it.