How to Make AI Remember You
The fastest way to make any AI tool more useful is to stop introducing yourself to it every session
The fastest way to make any AI tool more useful is to stop introducing yourself to it every session. There are several ways to do that — some built into the tools, some you set up yourself, and one that works across all of them at once.
Here's the full range, from quickest to most durable.
How do you make AI remember you?
Three levels, increasing in durability:
- Use the built-in memory features — quickest, but per-platform and reactive
- Maintain your own context file — more control, but manual and hard to keep current
- Use a portable context layer — set up once, works across every tool
Most people stop at level one and never realize the other two exist. Each solves a problem the previous one leaves open.
How do you set up memory in ChatGPT and Claude?
Start here — it's built in and takes minutes. Note that memory is often already on but never configured, so it's worth checking rather than assuming.
ChatGPT:
- Set Custom Instructions (Settings → Personalization → Custom Instructions): your role, what you work on, how you want responses. This is the highest-leverage field most users leave blank.
- Check Memory is enabled in the same panel. ChatGPT accumulates memories automatically, but reviewing and curating them keeps the picture accurate.
Claude:
- Claude now offers memory in supported accounts (including a free-tier option as of 2026); availability and exact menus vary by plan and rollout. Check Settings → Memory to see what's stored, edit it, or enable it.
- Set your Profile preferences (Settings) — the global instructions that load into every conversation.
This gets each tool to remember you within itself. The limit: what you set in ChatGPT does nothing for Claude, and vice versa.
Should you keep your own context file?
It's the next step up in control, and a real option people use: write a Markdown file about yourself — role, background, projects, preferences — and paste it into each new chat or attach it to projects. Some people keep it in a GitHub repo and pull from there.
The upside: you control exactly what's in it, and it works in any tool that accepts text. The downside is maintenance. When your role changes or a project wraps, you have to remember to update the file — and you're still manually pasting it into every tool, every time. In practice, the file drifts out of date within weeks, because updating it is a chore with no prompt to do it. It's better than nothing, and worse than something automatic.
Why isn't turning on memory enough?
Because every option above is either siloed, manual, or both.
Platform memory is reactive and locked to one app — three tools means three separate, partial memories. A hand-maintained file is portable but goes stale and still needs pasting. Neither gives you one accurate, current picture of yourself that every tool just has.
There's also a trap worth naming: connecting your tools (Gmail, Notion) to an AI via MCP and expecting it to pull your context. A connector gives the AI a way to search or retrieve from a source — useful, but not the same as a pre-extracted, structured profile of you. Depending on how retrieval works, asking for "everything about my Q3 strategy" can surface what literally matches and miss the document titled differently. Connecting a source isn't the same as having the right context ready at session start.
How do you make every AI tool know you?
Maintain your context once, in a layer that delivers it everywhere.
A portable context layer extracts a structured profile from your real sources — LinkedIn, Notion, email, GitHub — keeps it current as those sources change, and serves it to any MCP-compatible tool. You set it up once. Then Claude, Cursor, ChatGPT, and the next tool you try all start from the same accurate, up-to-date picture of who you are — without you configuring memory in each, or pasting a file that's three months out of date.
That's the difference between making one tool remember you and making AI, in general, know you.
→ How to build that context: How to Build Personal Context for AI
→ How it reaches your tools: How to Deliver Personal Context to AI Tools
→ Set up your context vault with Unabyss →
