The Conversation Mindset
Module 1, Reverse Prompting | Essay 2 of 8
There's a reason "just Google it" became a cultural shorthand. Search is fast, transactional and anonymous. You type a few words, you get links back, nobody needs to explain anything to anyone.
That habit is the enemy of good AI work.
When people first use an AI agent, they compress everything into the shortest possible request. "Summarize this." "Write an email." "Make a plan." It's search engine behavior applied to a completely different kind of tool. And it produces search engine quality results: technically responsive, utterly generic.
The mental shift that changes everything is simple. Stop treating your agent like a search engine. Start treating it like a smart colleague you're bringing up to speed.
Think about how you'd actually brief a capable new hire on a task. You wouldn't hand them a three-word instruction and walk away. You'd give them background. You'd explain why the task matters. You'd tell them who the audience is, what good looks like, what to avoid. You'd answer their questions. And you'd expect a first draft that needs refinement, not a finished product.
That's the conversation mindset. The agent is capable. It just needs the same thing a smart person needs: context about the situation, clarity about what you want, and room to ask questions.
What makes this hard at first is that it feels slower. Writing a three-sentence brief feels like more work than firing off a quick prompt. And in the short term, it's. But the output you get from a well-briefed agent is faster to use than the output you get from a vague one. Editing something that's 70% right takes five minutes. Starting from scratch because the first result missed entirely takes 20.
The other thing that shifts with this mindset: you stop needing the perfect prompt. There's a whole industry of "prompt engineering" built on the idea that if you just find the right magic words, the AI performs. That's largely a distraction. What actually works is having a conversation. Give context. Read the response. Tell it what's right and what's off. Keep going.
You don't need to be technical. You don't need to learn special syntax. You just need to communicate the way you'd communicate with anyone doing work on your behalf.
The agent is ready. It's been ready. It just needs you to talk to it.
Pod Exercise: Pick a real task from your work this week. Before you type anything in your practice pod, write two sentences: what the task is and why it matters. Then write two more: who the output is for and what good looks like. Use that as your opening message instead of a one-line command. Notice the difference in what comes back.