Memory you can see — and that doesn't drop what you tell it

A memory you can't see is a memory you can't trust. Two changes shipping today make yours both visible and reliable.

one-chat.app — same memory, any AI
GPTClaudeGeminimemory onswitching AIs — memory stays
Book somewhere nice for Mum's birthday dinner.
remembered · 3 things
She's vegetarian, loves being near the water, and her birthday is the 14th — here are three waterfront places with standout vegetarian menus, all free that evening.
Swap the AI above — it still knows all of this.

See what a reply remembered

Assistant turns now show a small "used N memories" marker when the answer drew on what the assistant knows about you. It's a quiet confirmation that your context is actually being used — not silently ignored — and a starting point when you want to check or correct what it's working from.

Writes that don't get dropped

Previously, saving a new memory happened in the background as your reply finished — and on a busy moment that write could occasionally be cut short, so the assistant seemed to "forget" something you'd just told it. Memory writes are now durable: they're queued and retried until they land, so what you share sticks.

You'll also see a subtle Indexing state while a new memory is being woven in, and a Saved confirmation once it's stored — so there's no guessing whether it took.

Still yours to edit

Everything the assistant remembers stays visible and editable in your memory view — correct a fact, remove one, or add context directly. Memory should feel like something you own and can inspect, not a black box that quietly shapes every answer.