When I know what I need
For a specific task, I try to give the assistant an outcome, source material and boundaries. For example: “Shorten this to one page. Keep the argument and the example. Don’t add new facts.” A clear brief makes the result easier to check.
But not every conversation with AI starts with a brief. Sometimes I only have a sense that something isn’t right. That can be a starting point too.
When I don’t yet understand the problem
For example: “I have a plan for the next month, but I keep putting off starting. I don’t know whether I’m overloaded, the first step is unclear, or the goal no longer makes sense. Help me separate these possibilities before we make another plan.”
This is an example, not a formula. The important distinction is between what I notice and how I explain it. AI can suggest questions and alternative interpretations. But it has no direct access to my experience, and a convincing answer is not evidence on its own.
Clarity can emerge through conversation
In LifeOS, I use AI to sharpen the question, not just answer it. I might begin with a fragment of thought, a voice note or feedback on an earlier result. Then I ask: what do we know, what are we assuming, and what is missing?
This work needs the relevant context. I don’t assume every chat remembers previous conversations or can access my documents. I check what it can actually use and share only the material it needs.
Return to what happened
Not every session needs to end in a plan. A better question, or a decision to leave things as they are for now, can be enough. When I do choose an action, I return to what actually happened. That helps me distinguish an answer that sounds good from one that was useful.