You had a great session yesterday. You explained your project, your preferences, how you like answers structured. Today you open a new chat and type the same thing from scratch. Again.
This is one of the most common frustrations with AI assistants — and it’s largely solvable.
The problem
Every new chat session starts blank. The AI has no idea who you are, what you’re working on, or that you prefer short answers without bullet points. So you either repeat yourself every time, or you get generic responses that miss the point.
The solution
- Open your AI assistant’s settings and find the memory or personalisation section
- Add a short profile: who you are, what you do, how you like answers formatted
- Tell it your preferences explicitly: timezone, units, tone, output format
- For ChatGPT — use saved memories (free) or let it learn from chat history (Plus/Pro)
- For Gemini — enable personalisation in settings; Pro users can also import memory from other tools
- For Claude — use Projects with custom instructions, or enable chat memory (available on all plans)
- After each useful session, tell it: “Remember that I prefer X” — it takes two seconds
How it works
Memory is not magic — it’s injected context
When you start a new chat, the AI doesn’t actually “remember” anything the way a person does. What happens is simpler: a summary of saved details gets quietly added to the start of your conversation. You don’t see it, but the model does. That’s why it feels like it knows you.
This means the quality of memory depends entirely on what’s in that summary. Vague entries like “user is a developer” produce vague personalisation. Specific entries like “I’m an engineering lead working in TypeScript, I want answers under 5 sentences, no bullet points unless I ask” produce noticeably better results.
ChatGPT has two layers
Saved memories are things you explicitly tell it to remember — either by saying “remember this” or by going to Settings → Personalisation and adding them manually. Chat history is a separate feature (Plus and Pro only) where ChatGPT pulls relevant context from your past conversations automatically.
If you’re on the free plan, saved memories are your main tool. Use them. A few well-written entries make a real difference.
Gemini works similarly, with one interesting addition
As of early 2026, Gemini lets you import memory from other AI tools — including ChatGPT and Claude. If you’ve already built up a good memory profile somewhere else, you can bring it over without starting from scratch. Go to Settings and look for the memory import option.
Claude takes a slightly different approach
Rather than a single global memory, Claude uses Projects — dedicated spaces where you write custom instructions once and they apply to every conversation inside that project. Think of it as a briefing document the AI reads before every session. You can describe who you are, what the project is about, your preferred tone, and any constraints.
Chat memory is also available on all plans as of 2026, and like Gemini, Claude supports importing memory from other tools if you’re switching over.
A word on privacy
Giving an AI assistant a detailed personal profile is a real trade-off. These summaries are stored by the provider and used to personalise your experience. Most platforms let you view, edit, or delete what’s saved. It’s worth checking what’s in there periodically — especially if you’ve been using it for a while and things have changed.
Wrapping up
Setting up memory takes ten minutes. After that, you stop re-introducing yourself at the start of every session. It won’t make the AI understand you perfectly, but it removes a friction point that adds up fast.