These are the five prompts from my reel, written out in full so you can copy and paste them. Ask them in order, in one chat, before you write a single resolution for 2027. Each one has a follow-up prompt, and there is a backup prompt if ChatGPT does not know you yet.

The idea is simple. Most resolutions fail because we pick goals without looking at ourselves first. If you have used ChatGPT for a while with Memory on, it may already remember a fair amount about what you ask for, what you worry about and what you keep putting off. These prompts ask it to read that back to you.

The short version

• Five prompts, asked one after another in the same chat: who you seem to be vs who you are, the insecurity you do not admit, the habit keeping you average, two versions of your life one year from now, and the one thing to do in the next 30 days.
• They work best when ChatGPT Memory is on, because the answers come from your past chats. Check it in Settings, then Personalization, then Memory.
• If Memory is off, or your account is new, use the backup prompt in this post. It makes ChatGPT interview you first.
• ChatGPT only knows what you have typed into it. Treat the answers as a mirror to think with, not as a verdict on you.
• Set aside roughly 20 minutes, more if you need the backup interview. Do it alone, and answer the follow-ups honestly.

Before you start: check that Memory is on

Memory is the ChatGPT feature that lets it remember things about you across chats. OpenAI’s help page lists several sources Memory can use (past chats, saved memories, custom instructions, saved files and connected apps such as Gmail, depending on your plan and region). Two matter for these prompts:

• Saved memories: details you asked ChatGPT to remember, or that it saved as useful context.
• Reference chat history: lets ChatGPT use relevant information from your past conversations.

For these prompts, chat history matters most. To check your settings:

1. Open Settings. OpenAI’s steps: on the website, open your account menu and select Settings. In the iPhone or Android app, open the sidebar and select your profile icon.
2. Select Personalization.
3. Select Memory.
4. Turn on the memory options you see. Depending on your account these may be called Memory, Reference saved memories or Reference chat history. OpenAI notes that if Reference saved memories is off, Reference chat history is off too.

OpenAI says memory features and controls “can vary by plan, region, platform, and workspace settings”, so your screen may not match the steps above. On the ChatGPT pricing page (checked 11 November 2026), the Free plan lists “Limited memory and context” and marks “Memory with past chats” as Limited. Go lists “Longer memory”, Plus “Expanded memory and context” and Pro “Maximum memory and context”. So the prompts should still work on the free plan, but I have not tested how much a free account remembers, and paid plans have more to draw on. Plan wording can differ by country.

One more thing to know: OpenAI says Memory “does not retain every detail from every conversation”. ChatGPT picks what it thinks is relevant. That is why the first step below is a quick test.

Step 0: test what ChatGPT knows about you

Open a new, normal chat (not a temporary chat) and paste this:

Before we start, tell me what you currently remember about me. Use your saved memories and our past chats. List it as short bullet points: my work, my goals, my routines, the problems I keep bringing to you, and how I usually talk to you. If you remember very little, say so plainly.

If you get a real list, carry on to Prompt 1 in the same chat. If the list is thin or wrong, use the backup prompt first.

Backup: if Memory is off or ChatGPT barely knows you

I want you to give me an honest read on myself, but you do not know me well yet. Interview me first.

Ask me 12 questions, one at a time, and wait for my answer each time. Cover: my work or studies, money, health, relationships, how I spend a normal weekday, what I wanted to do this year and did not do, what I avoid, what I am proud of, and what I am afraid of.

Do not give any advice during the interview. After the 12th answer, write a short summary of who I am based only on what I said, and ask me to correct anything that is wrong.

Answer properly, correct the summary, then continue with the five prompts in that same chat. Before Prompt 1, paste this one line so ChatGPT knows what to use:

For the rest of this chat, whenever I say "our past chats" or "your memory", use only my interview answers above. Do not tell me you lack access to past chats.

---

The 5 prompts

Keep everything in one chat. Each prompt builds on the answer before it, and prompts 4 and 5 will not make sense in a fresh chat.

Prompt 1: who the world sees vs who you really are

This is the warm-up. It shows the gap between the image you present and the patterns in what you actually ask about.

Based on everything you know about me from our past chats and your memory, answer two things.

1. Who do I appear to be to the world? Describe the image I try to present to other people.
2. Who am I really? Describe the person the patterns in our chats show: what I actually care about, what I worry about, and what I keep avoiding.

Then tell me the biggest gap between the two.

Rules: use specific examples from what I have told you. Do not flatter me. If you do not have enough information for a point, say so instead of guessing.

Follow-up:

Which parts of that are you confident about, and which parts are guesses? For each confident point, show me what I said that led you to it.

Prompt 2: polite mode off, the insecurity you never admit

This is the one I said would sting. ChatGPT tends to be agreeable, so you have to ask it directly to stop.

Now stop being polite. For this answer, do not soften anything and do not add encouragement.

From the way I write, the questions I ask and the things I keep coming back to, what is the one insecurity I have that I never admit, not even to myself?

Tell me:
- what the insecurity is, in one sentence
- the evidence for it from our chats
- how it shows up in my decisions
- what it has already cost me

Be direct. If you are not sure, tell me how sure you are.

Follow-up:

What do I do to hide this insecurity from other people, and what do I tell myself so I do not have to look at it?

Prompt 3: the habit quietly keeping you average

Here you want one habit only. A list of ten is easy to nod at and ignore.

Based on everything so far, name the one habit that is quietly keeping me average.

I do not mean an obvious bad habit. I mean something I do regularly that looks harmless, or even productive, but keeps me at the same level year after year.

Tell me:
- the habit, in one line
- when and how it usually shows up
- why I keep doing it (what I get out of it)
- how it connects to the insecurity you just described

Give me only one habit, the one that matters most.

Follow-up:

What triggers this habit, and what is the smallest replacement I could do in that exact moment instead?

Prompt 4: two roads, one year later

Now it gets practical. You see the same year twice: once where you fix the habit, once where you do not.

Show me two roads, both starting today.

Road A: I fix this habit starting today. Describe my life one year from now.
Road B: I change nothing. Describe my life one year from now.

For each road, cover my work and money, my health and energy, my relationships, and how I feel about myself.

Keep it realistic. No fantasy outcomes on Road A and no disaster on Road B, just the most likely result of each. Write both in second person, as if you are describing a normal weekday in my life.

Follow-up:

On Road A, what is the hardest moment in the first two weeks, and what exactly should I do when it arrives?

Prompt 5: the one thing for the next 30 days

This replaces the resolution list. One action, small enough to do on a bad day.

Now give me only one thing to do in the next 30 days. Not a list, one thing.

It should be the single action that moves me from Road B to Road A. Make it:
- specific: what, when, where and for how long
- small enough that I can do it on my worst day
- measurable, so I can answer yes or no each day

Then tell me what I should stop doing to make room for it, and how I will know on day 30 that it worked.

Follow-up:

Turn that into a 30-day plan in a table: week, daily action, and a one-line check-in question for each Sunday. Then write one sentence I can read every morning.

---

After the five answers: write your resolutions

Only now write your 2027 resolutions. Keep the chat open and paste this:

Using everything in this chat, help me write no more than 3 resolutions for 2027. Each one must be linked to the habit and the one thing above, have a clear weekly action, and have a way to measure it. Tell me which usual resolutions I should NOT make this year, and why.

If you want ChatGPT to check on you later, come back to this same chat on day 30, replace the text in square brackets with what really happened, and send:

It is day 30. Here is what I actually did: [WRITE WHAT HAPPENED, HONESTLY]. Compare it with the plan. Tell me what worked, where I slipped back into the old habit, and the one thing for the next 30 days.

How to get honest answers

• Use a normal chat. A temporary chat set to Unpersonalized does not use memory at all, so ChatGPT will have nothing to work with. A temporary chat set to Personalized can use your existing memories, but it has trade-offs. See the Privacy section below.
• Ask for evidence. If an answer sounds like a horoscope that fits anyone, reply: “That is too general. Give me examples from what I have actually told you.”
• Correct it. If it gets a fact about you wrong, say so and ask it to redo the answer. OpenAI’s help page confirms you can tell ChatGPT that remembered information is incorrect or outdated.
• Check your Custom Instructions. If you once told ChatGPT to always be positive or brief, that can blunt the answers. They sit under Settings, then Personalization on web and desktop, and under Settings, then Customize ChatGPT in the phone apps.
• Do not argue with every point. Read the answer, wait a few minutes, then decide what is true.

What this can and cannot do

It can	It cannot
Spot patterns in what you have typed over many chats	Know anything you never told it, such as your offline life
Say things a friend might be too polite to say	Be sure it is right. It can sound confident and still be wrong
Turn a vague wish into one small daily action	Do the action for you, or replace a doctor or therapist

If you mostly use ChatGPT for work, expect answers that lean heavily on work. The backup interview prompt fixes that by adding the rest of your life.

If the answers bring up something heavy, such as constant low mood or anxiety, please talk to someone you trust or a qualified mental health professional. A chatbot is not the right tool for that.

Privacy: read this before you pour your heart out

• Training. On personal accounts, OpenAI says it may use chats and remembered information to improve its models when “Improve the model for everyone” is on. OpenAI says this is on by default on personal Free, Plus and Pro accounts. The page does not name Go, so check the switch whichever plan you are on. To switch it off: Settings, then Data controls, then Improve the model for everyone. Your chat history stays. Even with it off, do not press thumbs up or thumbs down in this chat: OpenAI says the whole conversation attached to that feedback may still be used for training. Details are on the Data controls help page.
• This chat can become memory. With Memory on, what you say here may be used to personalise later chats. You can review it in Settings, then Personalization, then Memory, or just ask ChatGPT what it remembers. If you switch Reference chat history off later, OpenAI says what it remembered from past chats is scheduled for deletion within 30 days. The chats themselves stay in your history.
• Connected apps and files. If you have connected apps such as Gmail, or saved files in ChatGPT, Memory may draw on those too when you ask for “everything you know about me”.
• Deleting. OpenAI says that to fully remove something, you should delete both the saved memory and the chat where you shared it. Deleting only the chat does not always delete a saved memory made from it. OpenAI says deletion can take time, and it may keep logs of deleted saved memories for up to 30 days.
• Temporary chat. A temporary chat stays out of your history, does not create memories and is not used for training while it stays temporary. OpenAI may keep a copy for up to 30 days for safety. If you choose Personalized when you start it, it can use your existing memories, so you can run the five prompts there if you do not want this conversation kept. Two trade-offs: the answers may be thinner than in a normal chat, and you cannot reopen a temporary chat once you leave it. Copy your one thing and the 30-day plan somewhere first. The day-30 check-in only works if you save the chat, which turns it into a regular chat that follows your normal memory and training settings. If you want nothing here to use or create memories, start an Unpersonalized temporary chat and use the backup interview.
• Shared devices. If someone else uses your phone or laptop, they can open this chat. Delete or archive it if that worries you.
• Leave out passwords, bank details, ID numbers and other people’s private information. The prompts do not need them.

Common problems

ChatGPT says it does not have memory of past chats

Check Settings, then Personalization, then Memory, and make sure you are not in a temporary chat. If the options are not there, OpenAI says they may not be available for your account yet. Use the backup interview prompt instead.

The answers are too soft

Reply with this:

You are still being too gentle. Rewrite that answer as if you were a blunt mentor who has watched me for a year and has nothing to gain from being nice. Keep it accurate, and keep the evidence.

The answers are harsh but feel wrong

Ask for the evidence. If it cannot point to something you actually said, treat that part as a guess and ignore it.

Can I do this in Hindi or Hinglish?

Yes. Add this line to the end of any prompt:

Reply in simple Hinglish (Hindi written in English letters).

Does it work in other AI apps?

The backup interview version works in any chatbot, because you supply the information yourself. The memory-based version depends on how much that app remembers about you. I wrote these prompts for ChatGPT and have not checked them in other apps.

Sources

• Memory in ChatGPT, OpenAI Help Center: what Memory is, the settings path, Reference chat history, saved memories, deletion and model training.
• Temporary chat in ChatGPT, OpenAI Help Center: Personalized vs Unpersonalized, no new memories, 30-day safety copy.
• Data controls in ChatGPT, OpenAI Help Center: the Improve the model for everyone setting, where to find it, and the note on feedback.
• ChatGPT Custom Instructions, OpenAI Help Center: where custom instructions live on web and mobile.
• ChatGPT plans and pricing: memory wording for Free, Go, Plus and Pro.

All pages checked on 11 November 2026. OpenAI changes settings names often, so if a label looks different in your app, look for the closest match under Personalization. The prompts are my own wording. This post is general information, not medical or psychological advice.