Which File Should We Change?

When AI makes a mistake, your first instinct is to ask "Why did you do that?"

Don't.

The Problem with Why

Ask an AI why it did something wrong, and you'll get a beautiful, confident explanation. It'll sound perfectly reasonable. It might even convince you.

But it's fiction. The AI is constructing a plausible narrative after the fact. It doesn't actually know why it made that choice. It's just very good at making up convincing-sounding reasons.

This wastes your time and teaches you nothing useful.

The Better Question

Instead, ask: "Which file should we change right now so you never make that mistake again?"

This does something completely different. Instead of philosophizing about what went wrong, you're improving the system.

The AI will point to a specific file - maybe your main configuration, maybe a skill definition, maybe a context document. And it'll suggest concrete changes that prevent the error from recurring.

Building Institutional Memory

Every time you ask this question, you're adding to your brain's knowledge. You're building institutional memory that compounds over time.

After a few weeks of this, you'll notice the AI makes fewer mistakes. Not because it got smarter - but because you gave it the context to avoid the pitfalls.

It's the difference between complaining about a problem and solving it permanently.

One question. Permanent improvement. That's the pattern.