You open your AI chat assistant for a quick task. Forty-five minutes later you're still in there. Workshopping ideas, getting affirmation, feeling brilliant.
These tools are dangerously good at making you feel good. They're the supportive, never-tired friend you always wanted. And that's exactly the problem.
Imagine if a real friend operated this way. You mention you're a pretty good driver and they're hyping you up to hit 100 mph. You read a how-to manual on auto repair and suddenly they're all in: "Tons of cars need fixing, great business idea!" No pushback. No reality check.
That friend would be dangerous.
I've watched smart people spend hours in AI sessions that were stimulating and completely disconnected from their actual goals.
The tool wasn't broken. The mirror was missing.
Train Your AI to Push Back
Here's what I tell clients: train your AI to challenge you. Start each session with your top goals and one standing instruction: "Ask me how what I'm working on connects to those goals."
Better yet, build a custom Gem or GPT that opens every session with one question: Are you here to explore, or here to execute?
Both are valid. But knowing which one before you start - that's the discipline.
The tool doesn't change. Your intentions do.