August 19, 2026

Being nice feels better

There’s a surprising side-effect of using an AI agent to do stuff for me.

When it does something “dumb” I lose my mind.

For example...

Last week I spent a lot of time going back and forth with Hobbes (my openclaw) about how to generate audiograms from my podcast episodes.

After an hour or two, we arrived at a solution that I liked. I told it to lock this in as a skill so next time there would be no discussion and it would just create the clips to spec.

Fast forward to today.

I gave it a new episode to process and it did A BUNCH OF STUFF that I had PAINSTAKINGLY corrected the last time.

Like... it made the EXACT same mistakes that I had EXPLICITLY corrected.

🤬🤬🤬🤬🤬

I’ve never had employees, but I’m sure the same situation with an intern or VA forgetting what they had learned the previous week would be frustrating.

However, I’m positive that I would have way WAY more patience with a human than I do with Hobbes.

When Hobbes forgets its training, I go straight from chill to THERMONUCLEAR

You might think, “Well, who cares if you yell at a bot?”

And the answer is that even though I have hurt nobody’s feelings by yelling at a computer, I hate the feeling of acting like that.

Even days later, if I happen to scroll back and see one of my ALL CAPS outbursts, it pulls me right back into the gross feeling of behaving like an unhinged maniac worst-boss-of-all-time jerk.

So, note to self:

I should be nice to my AI, not because I think it has feelings, but because I don’t like the feeling of not being nice.

🤔