OpenAI Made ChatGPT's Memory Better. It Still Isn't Yours.

OpenAI just shipped a real upgrade to ChatGPT's memory. It picks up more of what you tell it, keeps up as things change, and finally shows you a page of what it knows about you. If you live in ChatGPT, you'll feel the difference. Use it.
One thing the launch leaves out: all of that memory lives inside ChatGPT, and it stays there. That's the whole case for what we're building. OpenAI just spent a launch convincing everyone that an AI remembering you is worth having. They left out a word: whose.
The better the memory, the taller the wall
Think about what you tell an AI over a year. What you are working on. How you like things explained. The project you have been buried in for months. The preferences you would have to repeat to a brand new assistant on day one. ChatGPT now remembers a lot of it. And all of it stays inside one company's product.
So the day a better model arrives from someone else, and one always does, you start from zero. New assistant, blank slate, nice to meet you. Every preference and every piece of context, taught again by hand.
That is the quiet trade every AI company is offering right now: we will remember you, as long as you never leave. The better the memory, the taller the wall.
The memory that comes with you
All of that is your work, and it's sitting in a product you don't own. We think that's backwards. Your memory is the one piece of this that's actually about you, not about whichever model is ahead this month. It should move with you.
That's what Tabula is. Save something once and it's there in ChatGPT, Claude, Mistral, Grok, Perplexity, Notion, whatever you open today and whatever's best next year. Connect it once, switch models without starting over, run three at the same time and have them all know the same things about you.
And because it's yours, you control it. Read everything your AIs have saved about you, fix what's wrong, delete what you don't want, and export all of it whenever you like. ChatGPT shows you its memory now too, which is good. What it can't do is let you carry that memory to the next AI. For Tabula, that's the whole point.
OpenAI's update is good news for us, honestly. The whole world is about to find out how useful it is when an AI remembers you. We just hope people learn the second half right after the first: a memory you can't take with you isn't really yours, no matter how good it gets.
Questions
- Is ChatGPT's new memory bad?
- No, it's a real improvement. The catch is it only works inside ChatGPT. The moment you switch to another AI, none of it follows you. That's the gap Tabula fills: one memory that works across every assistant you use.
- How is Tabula different from ChatGPT's memory?
- ChatGPT's memory only works inside ChatGPT. Tabula is one memory that works across every AI you connect (ChatGPT, Claude, Mistral, Grok, Perplexity, Notion), and you can read, edit, export, or delete all of it whenever you want.
- Can I use Tabula and ChatGPT's memory at the same time?
- Yes, they're not mutually exclusive. ChatGPT keeps its own memory, and Tabula gives you a portable copy that also works in every other AI you open.
- Does Tabula read my conversations in the background?
- No. Tabula keeps what you or your assistant decide is worth saving, not everything you say. You choose what goes in, and you can see or remove anything at any time.