🗂️ System: build a context profile so you stop re-explaining yourself

You did good work in a chat last week. Today you reach for it and it’s gone. Or the chat got long and forgot what it knew an hour ago. And every new chat, you retype who you are and how you want the answer, because the last one forgot.

Your AI work feels messy because every new chat starts from zero. The context never got a place to live, so you keep rebuilding it.

One of you put it simply: “context loss kills flow more than people realize.”

You don’t have to write your profile from a blank page. The model pulls it out of you. Paste this into ChatGPT:

I want to create a reusable context profile for my work so I stop
re-explaining myself in every chat.

Don't write the profile yet. First, interview me one section at a time,
wait for each answer before the next question. Cover: my role, who I serve,
my current main goal, the tools I use, the tasks I keep coming back to, my
tone, my decision style, and what a genuinely useful answer looks like for me.

When we're done, write TWO versions:
1. A short version (5-6 lines) I can paste at the top of any normal chat.
2. A full version I can save and upload into a Project.

Include a "don't assume" section and a "verify before answering" section.
Keep it practical, cut the fluff, and don't add personal details I didn't give you.

What you’re doing

  1. Run the interview. Answer like you’re talking, not writing a resume. About 5 minutes.
  2. Create a Project called My Work Context (free in both ChatGPT and Claude). Upload the full version.
  3. Next task you’d normally open a blank chat for, open the Project instead. It already knows you.

One honest caveat: the profile isn’t a secret-keeper. Keep it to working context, no client-confidential details. And update it when your work changes; a stale profile makes answers worse without telling you.

If your context profile is working and you want to turn it into a reusable setup, next move: :rocket: Share: turn one working setup into a template you can hand off

Reply with what happened

Run the interview, then reply:

  • Post the first line of your context profile.
  • What did it get right about you, and what did you have to correct?

Here’s mine, to show the shape: “I write for non-technical professionals who already use tools like ChatGPT, Gemini, and Claude at work but feel they’re only getting a fraction of the value.”

That first line is usually the one that changes how every answer comes back. Borrow good ones from each other.