My wife told me I'm communicating better. Clearer, fuller, easier to follow. And the reason — the part that still makes me laugh — is AI prompting.
I didn't set out to become a better partner. I set out to get better answers from a machine. The first was a side effect of the second.
The fact I'd been ignoring
Here's what prompting forced me to confront: my mind cannot be read.
Obvious, right? Except I lived as if it weren't. I'd hand someone half an idea and expect the whole thing to land — because they know me, because we have history, because surely the context was already there.
It wasn't. I was assuming it was.
(Remember that word. Assuming.)
Why we assume — and why it's a trap
The assumption comes from memory. The other person knows me, so I trust the context is already loaded. And sometimes it is. Think of your oldest friend — you don't explain everything. Half a sentence and they're already nodding.
But that's the exception, not the rule. Most of the time the context isn't shared. It only feels shared because it's vivid in your own head.
Assuming is a bet. And it's a bigger one than we realize.
AI doesn't pretend to know you
This is where AI taught me the lesson cleanly — because AI never fakes the context.
When the context isn't clear, AI doesn't quietly guess and hope you won't notice. It assumes whatever its own knowledge suggests — and you watch, in real time, the gap between what you meant and what you actually said. The result is only as good as the clarity you gave it.
Talking to AI is like meeting someone for the first or second time. There's no history to lean on. You have to spell things out — the goal, the constraints, the part you think is too obvious to mention. Especially that part.
There are ways to give AI a persistent memory so you stop repeating yourself. But that's a story for another article.
The skill bled back into my life
Once I started naming the unspoken — the context, the intent, the thing I assumed was understood — my prompts got better. Then my conversations did too.
The discipline turned out to be the same. Say the quiet part. Don't make the other person reconstruct what's in your head. Close the gap instead of betting they'll cross it for you.
I went looking for better outputs from a tool. What I found was being understood — by my wife, by a client, by a colleague reading an email.
That's the trade I didn't see coming. And it's the one I'd keep.
The fix was never better words. It was remembering that no one can read your mind — not the AI, not the person across from you. So try to stop assuming they can. Say the part you think is too obvious to say. And see how it goes.
Worked with Claude (Anthropic) as thinking partner · Jun 2026