Breaking Big Requests into Conversations
Module 1, Reverse Prompting | Essay 7 of 8
There's a temptation with AI agents to hand them the whole problem at once. One prompt to cover everything. Research, plan, draft, format, do it all.
The output from that prompt is almost always disappointing. Not because the agent can't handle complexity, but because complexity requires stages, and stages require conversation.
Think about how a project actually unfolds. Research happens before writing. Writing happens before editing. Editing happens before formatting. Each stage produces something that informs the next. Collapsing all of that into a single prompt skips the moments where you review, redirect and add new information, the moments that make the output actually right for your situation.
Breaking work into conversations changes this. Instead of one sprawling prompt, you have a series of focused exchanges. Each one has a clear goal. Each one builds on what came before.
Here's what that looks like in practice. Say you need to prepare for an important client meeting. Instead of: "Help me prepare for a client meeting next Tuesday about our Q2 performance", try three conversations:
First: "Help me identify the three things this client cares most about based on what I tell you about them." Share what you know. Get a focused list.
Second: "Here's our Q2 performance data. Given those three priorities, what's the most important story to tell?" Get a narrative, not a data dump.
Third: "Help me structure a 20-minute agenda that leads with that story and leaves 10 minutes for questions." Get something you can actually use.
Each conversation is manageable. Each output feeds the next. By the end you have something built through real back-and-forth, not something generated in one shot and hoping it covered everything.
The natural break points are usually: before you need new information from outside, before the output changes form (research to draft, draft to edit), and before the audience or purpose shifts.
You don't need a rigid system. Just the habit of asking: is this one task or several? If several, start with the first one.
Pod Exercise: Take a complex task you've been putting off, something that feels too big to tackle. Break it into three to five separate conversations, each with its own clear goal. Run the first conversation in your practice pod. Just the first one. Then use that output to brief the second.