How to Use ChatGPT the Jay Shetty Way: His 7 Reflection Prompts

If you searched “how to use ChatGPT Jay Shetty”, you are almost certainly looking for his On Purpose episode “7 ChatGPT Prompts That Will Actually Change Your Life” (published March 20, 2026). His core idea is to use ChatGPT for structured self-reflection rather than productivity: run a life audit, unpick self-sabotage, rehearse hard conversations, and more. All of it with one big caveat, which is that AI is a tool for self-awareness, not a replacement for real people.

Below is what the episode covers, how to set ChatGPT up for this kind of use, and where we think the approach works and where it doesn’t.

What Jay Shetty actually recommends

According to the official episode page, Shetty frames ChatGPT as a space for structured conversations with yourself instead of a way to draft emails or plan meals. The episode is organised into seven chapters:

A note on accuracy: the chapter titles come from the show notes. Partial transcripts and podcast summaries (for example on Podscripts) quote fragments of some prompts, but we could not verify the complete wording of all seven. The example prompts below are our own wording based on each chapter, unless they are marked as quoted. For Shetty’s exact phrasing, listen to the episode on his site, iHeart or YouTube.

Before you start: set ChatGPT up for personal reflection

These prompts invite you to share personal details, so spend two minutes on settings first. As of November 2026 these controls exist, but OpenAI moves menus around regularly.

1. Decide whether you want memory on. OpenAI’s memory announcement explains that you can turn off referencing saved memories or chat history in Settings. Memory helps with ongoing accountability (prompt #5). Turn it off if you’d rather nothing carries over.
2. Use Temporary Chat for the sensitive ones. OpenAI says temporary chats don’t appear in history, don’t use memory and aren’t used to train its models, though a copy may be kept for up to 30 days for safety purposes. That suits prompts #6 and #7.
3. Know that deleting a chat isn’t deleting a memory. Per OpenAI, if ChatGPT saved a memory during a chat, you have to delete the memory itself.
4. Leave out other people’s identifying details. When you describe a partner, parent or boss, a first name or “my manager” is enough.

The seven prompts, step by step

1. The brutally honest life audit

Shetty’s starting point is to rate each area of your life honestly, from health to relationships to career, then let ChatGPT find the patterns and root causes.

1. List 6 to 8 life areas (health, relationships, work, money, friendships, purpose, rest).
2. Give each a score out of 10 and one sentence on why.
3. Paste it in with a prompt like this: “Here is my honest self-rating across my life. Identify the patterns you see, the likely root causes behind the lowest scores, and the one area that would lift the others if I improved it. Ask me questions before giving advice.”

That last line matters. Asking ChatGPT to question you first gets you out of generic advice.

2. Decode your self-sabotage patterns

Summaries describe this one as reverse-engineering self-sabotage. Describe two or three times you got in your own way. Then try: “What do these situations have in common? What might I be protecting myself from, and what tends to trigger the pattern?”

3. Build a personal operating system

The idea is to pull core principles out of your own experiences. Share a few turning points, both good and bad, and ask ChatGPT to draft 5 to 7 personal principles in your own language. Then edit them until they actually sound like you.

4. Practise hard conversations first

Shetty suggests using ChatGPT as a rehearsal partner to practise your responses and get clear on the outcome you want.

1. Describe the situation and the result you’re hoping for.
2. Ask ChatGPT to play the other person realistically, including pushback.
3. After a few rounds, ask for feedback on your tone and on what you avoided saying.

5. Build an accountability system that works

Transcript excerpts quote this prompt asking ChatGPT to “call me out” when excuses appear and to “help me adjust the system, not abandon it” when you struggle. Give it your goal, your current habits and a check-in rhythm, and come back to the same chat (or keep memory on) so it has context.

6. Break down your emotional patterns in real time

According to a transcript excerpt, you describe a person or situation and ask ChatGPT to “reflect back to me the core emotion, and unmet needs underneath all of it.” This works best right after something has upset you, while the detail is fresh.

7. Write the letter you never sent

The chapter is about writing to someone you never got closure with. We couldn’t confirm Shetty’s exact prompt. A reasonable approach is to write the letter yourself and then ask ChatGPT to help you name what you’re really feeling, rather than having it write the letter for you. You don’t send it. Writing it is the whole point.

Where this approach works, and where it doesn’t

Strengths. ChatGPT is patient, available at 2 a.m., and good at spotting repeated themes in what you tell it. Shetty’s point is that it can show you blind spots that solo journaling tends to hide.

Limits. ChatGPT only knows what you tell it, and if you describe a conflict one-sidedly, its reflection will be one-sided too. It can also sound more confident than it should.

It is not therapy. Shetty says plainly that AI is not a replacement for human connection, and that the goal is self-awareness you then bring to the people around you. OpenAI’s own February 2026 update on mental health work says it is improving how ChatGPT detects distress and is training it to steer people toward real-world support with input from clinicians. If you’re struggling, contact a qualified professional or a local crisis line, not a chatbot.

Quick tips for better results

• Be specific. “I procrastinate” gets generic advice. “I put off replying to my sister for three weeks after she criticised my job” gets something useful.
• Ask it to challenge you. Add “push back if I’m rationalising” to any prompt.
• One topic per chat. It keeps the context clean and makes chats easy to delete.
• Act offline. The episode’s message is that the work happens in your real life. ChatGPT just helps you see it more clearly.