Article
Just because you can automate something doesn't mean you should.
For a long time, I had Astra call me every morning at 6:30. I would pick up and get a quick brief on what was happening and what I had that day.
The setup worked. Astra called. I answered. She gave me the information.
After a while, though, I realized the call wasn't valuable. The news was similar every day, and hearing, “Hey, here's your appointment,” did not improve my morning. It was faster to glance at my calendar.
Meanwhile, the call kept using tokens. So I killed it.
Measure the outcome
It is easy to judge an automation by whether it runs. The call happened at 6:30, so the technical part was doing its job.
That wasn't enough. The automation needed to serve me, and it wasn't. It used resources while adding a step between me and information I could get faster on my own.
The useful test is the outcome:
- Does the automation save time?
- Does it make the information more useful?
- Is it better than the simple option already available?
In this case, the simple option won. A glance at the calendar was faster than answering a phone call and listening to the same kind of update again.
There was no reason to keep the automation alive just because I had already built it.
Pay attention to where the process takes you
While I was explaining this on video, I had wandered down a path and had no idea where it was going. Then I walked into a spiderweb.
It was a pretty accurate picture of the automation.
I had started with a reasonable goal: wake up and get a useful brief. But I kept following the process without paying much attention to whether it was still serving me. Eventually, I ended up somewhere I did not need to be and had to figure out the way back.
Automation can do that. A process can keep running exactly as designed while producing very little value. If you only check whether it fired, completed, or delivered something, you can miss the more practical question: was it worth doing?
Kill the automation when the numbers do not work
The goal is to improve the outcome, even when that means retiring an automation you built.
Sometimes the right move is to adjust the process. Sometimes the right move is to shut it down and use the simpler method.
My morning call was working, but it was wasting tokens and taking longer than checking my calendar. That was enough information to make the decision.
If the simpler option serves you better, use it.