Pick the one job you do by hand every week. Not the biggest one. The one that keeps coming back.
Now put a clock on it. Time it start to finish, the next time you sit down to do it. Do not estimate, because your estimate will be wrong in a specific direction. Work you do on autopilot never arrives as a bill. It arrives four minutes here and twenty there, spread so thin across a week that it never looks like the thing to fix.
That number is the whole argument. Not a statistic about AI. Not a case study from a company that looks nothing like yours. Your number, for your job, measured once.
Last week I wrote about the conversations that go cold while you are busy. This is the same problem wearing different clothes. Both are cheap to ignore in any single week and expensive over a year, and neither one ever announces itself.
I found my own version of this by admitting it in front of you a few weeks ago. My problem was never building the thing. It was getting anyone to look at it once it existed. What fixed it was not motivation. It was putting the job on a repeating schedule with a date attached, the same way I would have handled it if it belonged to somebody else.
Here is why I keep pointing people at a weekly job instead of a big one.
It repeats, so whatever you fix pays you back every week rather than once. You already know every step, which means there is nothing to design, and design is where most of these attempts die. And it is small enough to finish in one sitting. That last part matters more than it sounds. A half-built automation is worse than none, because now you have a thing you do not trust and still have to check.
Most owners do the opposite. They go looking for one tool that will handle everything, and everything is too big to describe, so the tool gets configured for a week and abandoned. One repeating job fits in a paragraph. That is the entire difference between the two approaches, and it has nothing to do with which tool you picked.
One warning, because this is the part people get backwards. Do not automate the job you hate. Hate is a feeling and it is a bad selector. The job you hate might happen twice a year. Automate the job that repeats, even if you do not mind doing it, because repetition is arithmetic and arithmetic is the only thing here that compounds.
So: one job, one clock, this week. If the number surprises you, you have found your first one. If it does not, you have spent four minutes and can stop reading me on this subject.
If you want a faster way to see where AI could actually help in your business, skip the guessing. I built a free scan that reads your site the way an AI assistant would and shows you what it can and cannot work out about you. About a minute, nothing to install, no call: catalystworks.consulting/scan
Three things before you go
Forward this to one business owner who is trying to figure out where AI fits in their business. One forward is more useful to me than ten retweets.
Was this forwarded to you? Subscribe at signal.catalystworks.consulting. The moment you do, you also get a free scan that shows how visible your business is when people ask AI for a recommendation.
Reply with one sentence about what is stuck in your week. I read every one and it shapes what ships next.
Boubacar
I spent over a decade in HR at a Fortune 100. Now I'm a fractional AI leader who can help small businesses put AI to work.
[email protected] · signal.catalystworks.consulting