How to Use ChatGPT Memory: Make It Remember Your Preferences
Every time you start a new ChatGPT conversation, you used to start from zero. No context about who you are, how you like things written, what you do for work, or what you’ve already discussed.
ChatGPT Memory changes that. It lets ChatGPT retain information between conversations — so you stop re-explaining yourself every time.
This guide covers how it works, how to set it up, what gets stored (and what doesn’t), and how to control it.
What Is ChatGPT Memory?
Memory is a feature that allows ChatGPT to save facts about you across conversations. When you tell it something relevant — your job, your writing style preferences, your dietary restrictions — it can remember that and apply it automatically in future sessions.
Instead of starting every conversation with “I’m a marketing manager who writes for a B2B SaaS audience,” ChatGPT already knows.
Memory is available to ChatGPT Plus and Team subscribers. Free-tier users have limited or no access depending on rollout status in your region.
How to Enable Memory
- Go to ChatGPT and click your profile icon (bottom-left)
- Select Settings
- Navigate to Personalization
- Toggle Memory to On
That’s it. Once enabled, ChatGPT will start saving relevant information from your conversations automatically.
How Memory Actually Works
Memory isn’t a transcript of your chats. ChatGPT doesn’t record every word — it extracts and stores specific facts and preferences.
When you say something like “I prefer bullet points over long paragraphs” or “I’m vegetarian,” ChatGPT identifies this as a persistent preference worth remembering and adds it to your memory store.
You can see exactly what’s been saved at any time. Here’s how:
- Go to Settings → Personalization → Manage Memory
- A list of saved memory items appears
- You can delete individual items or clear everything
What Gets Remembered?
Memory stores things like:
- Personal facts — your name, job, location (if you share them)
- Preferences — writing style, formatting preferences, tone
- Context — ongoing projects, your business type, your audience
- Instructions — things you always want (or never want) ChatGPT to do
For example, after a few conversations, your memory might contain:
- “Prefers concise responses without filler phrases”
- “Works in real estate; often needs contract summaries”
- “Uses British spelling”
- “Runs a food blog focused on quick weeknight dinners”
Every future conversation gets that context automatically loaded in.
How to Manually Add Memories
You don’t have to wait for ChatGPT to notice things. You can tell it directly:
“Remember that I always want you to write in an active voice.” “Remember that my target audience is small business owners, not enterprise.” “Remember I’m gluten-free so any recipes you suggest should reflect that.”
ChatGPT will confirm it’s saved the information. You’ll see a small memory icon (a brain) appear when it’s actively drawing on stored context.
How to Edit or Delete Memories
Bad memories accumulate just like good ones. If ChatGPT stored something wrong or outdated, fix it:
- Go to Settings → Personalization → Manage Memory
- Find the item you want to change or remove
- Click the trash icon to delete it
You can also say in chat: “Forget that I work in finance — I switched to healthcare.” ChatGPT will update accordingly.
To wipe everything: Manage Memory → Clear All Memories. This resets your profile completely.
Temporarily Disabling Memory
Sometimes you want a conversation where nothing gets remembered — a sensitive topic, a hypothetical scenario, a creative experiment.
Use a Temporary Chat:
- Click the pencil/compose icon at the top-left
- Select Temporary Chat
- Nothing from this session will be saved to memory
Alternatively, turn Memory off entirely in Settings anytime. Existing memories are preserved but not used until you turn it back on.
Privacy Considerations
Here’s what you should know:
- Memories are tied to your OpenAI account
- OpenAI’s standard data policies apply — conversations may be used to improve models unless you opt out (Settings → Data Controls → Improve the model for everyone)
- If you share your account, other users could see your memories and add to them
- Enterprise accounts have stricter isolation; Team accounts keep memory per-user
Be thoughtful about what you share. Don’t put passwords, financial details, or sensitive personal data into memory. It’s not a secure vault — it’s a preference store.
Real-World Use Cases
For writers: “Remember I write in second person for how-to articles, keep intros under 50 words, and avoid the word ‘utilize.’”
For professionals: “Remember I’m a tax attorney and my clients are high-net-worth individuals. Adjust complexity accordingly.”
For productivity: “Remember I use the GTD framework for task management. When I describe a task, suggest where it fits.”
For health: “Remember I’m managing Type 2 diabetes — flag high-sugar options in any recipes or meal plans.”
For students: “Remember I’m studying for the LSAT and respond to logic questions with explanations of the reasoning, not just answers.”
The Payoff
Memory turns ChatGPT from a generic assistant into something that actually knows you. Over time, the friction of re-explaining your context disappears. Responses get more relevant. You spend less time prefacing every prompt.
It’s not perfect — ChatGPT still sometimes misses context or over-applies memories to situations where they don’t fit. But the improvement over cold-start conversations is significant.
If you’re a regular ChatGPT user, enabling memory is one of the highest-ROI things you can do. Set it up, populate it deliberately, and let it work.