Memory

With Long-Term Memory on, the assistant remembers facts about you and recalls them in later conversations. Tell it once that you use TypeScript and prefer terse answers, and you do not tell it again.

Turn it on or off under Settings → AI & Assistant → General → Long-Term Memory.

What gets remembered

Durable facts, not conversation transcripts — the sort of thing that would still be true next month. Preferences, the tools you work in, ongoing projects, people you mention often. Passing details of a single question are not worth remembering and generally are not.

Memories are only recalled when they are relevant to what you are currently asking, so an unrelated conversation is not coloured by everything the assistant has ever learned.

Reviewing what it knows

Settings → AI & Assistant → General → Manage Memories opens the memory manager. Every memory is listed and individually deletable, and there is a graph view showing how memories relate to each other — useful for spotting a cluster that has drifted out of date.

Edit or delete anything that is wrong. A wrong memory is worse than a missing one, because it will keep being applied.

Keeping it accurate

When the assistant learns something that overlaps with what it already knows, it updates the existing memory rather than storing a second, contradictory copy. That is why the list stays a readable size instead of growing forever.

Turning it off

Switching Long-Term Memory off stops both storing and recalling. Existing memories are kept, so switching it back on restores what was already there. To remove them for good, delete them in the memory manager.