Every conversation you have with ChatGPT leaves a trail. The model picks up on your writing style, your industry, your assumptions, your blind spots — and it quietly builds a working model of who you are. Most people never ask it to show its work. This prompt does exactly that.
This works because ChatGPT is pattern-matching constantly, even when you think you're just asking a one-off question. Surfacing those inferences gives you two things: a clearer picture of how the model is framing your requests, and a fast diagnostic for whether it has you pegged correctly — or has drifted into assumptions that are quietly skewing your outputs.
The Prompt
Based on everything I've shared with you in this conversation — and any patterns you've noticed in how I write, what I ask, and what I assume — tell me what you think you know about me.
Be specific. Include:
- What industry or role you think I'm in
- What level of expertise you're assuming when you answer me
- What goals or problems you think I'm working toward
- Any assumptions you've made about my audience, business model, or working style
- Anything you've inferred that I haven't stated directly
Then flag anything you might be getting wrong. Be honest, not flattering.When to use it
Pull this prompt out any time you want to audit the model's mental model of you — especially when outputs are starting to feel generic, misaligned, or like the model is writing for someone else.
- You've been in a long thread and the responses have gotten vague or off-base — run this to see what assumptions have stacked up.
- You're about to use ChatGPT for something high-stakes (a sales page, a pitch, a content strategy) and want to confirm it's calibrated before you start.
- You're onboarding ChatGPT to a new project and want to test whether your framing came through clearly in your first few messages.
- You're getting outputs that feel like they're written for a beginner when you're an expert — or vice versa — and you want to see exactly what the model thinks your level is.
- You want to build a better custom system prompt and need a starting point: the model's inferences give you a mirror to refine against.