At some point, I started using this tool with clear boundaries.
This is just for summaries.
This one’s for cleaning up meeting notes.
Drafting a plan? It works.
Final version? No.
In the beginning, GPT felt like it could do everything.
But the more I used it, the more I saw:
It’s good at some things—
and those few things, it’s very good at.
That’s when I began to divide the roles.
Tasks for people.
Things I have to do myself.
Jobs I can hand over to GPT.
I drew the line.
That was the moment I truly made AI a tool for myself.
When I used it without boundaries, it disappointed me.
Once I gave it limits, it became useful.
This wasn’t about emotion.
It was about judgment.
How far AI could be integrated—
that decision was always mine to make.
I no longer expect GPT to be brilliant.
Instead, I ask:
“Can this function be repeated?”
“Can this slot into my system?”
That’s all that matters now.
The moment cold filtering began,
GPT became a reliable tool in my hands.
I dropped the wrong expectations.
What remained was clear.
And that clarity—
that’s why I’ve kept it in my workflow.