The worst AI workflow mistake isn't a bad workflow. It's a workflow you have to
By John Aspinall ✱ · May 31, 2026 · Curated by George's Blog
The worst AI workflow mistake isn't a bad workflow.
It's a workflow you have to remember exists.
Sounds stupid.
But almost everyone building skills falls into this trap.
You build something that saves 30 minutes.
Awesome.
Then you realize the skill only works when you remember to run it.
So now you have a new job:
Remembering.
I built a skill that mines my recorded calls for LinkedIn content.
Genuinely useful.
Then one day I realized something.
I forgot to run it.
Again.
Think about what happened there.
I did the hard part.
I built the workflow.
I tested it.
I improved it.
And somehow I still created more work for myself.
Because now I have to:
□ Remember the skill exists
□ Remember when to run it
□ Remember why I built it
□ Feel guilty when I forget
At least manual work is honest.
You know you have to do it.
A skill that saves time but adds mental load is a half-finished job.
The fix is boring.
Automate the trigger.
If a workflow should run every day at 6 PM...
Run it every day at 6 PM.
If it should process every recorded call...
Trigger it when the recording appears.
If it should generate a report every Monday...
Schedule it.
The value was never the skill.
The value is never having to think about the skill again.
If you're still manually firing your AI workflows, you haven't eliminated work.
You've built yourself a second to-do list.
Finish the job.
Automate the run.