Skip to main content
Two memories. They do not share a list, and clearing one does not clear the other. With nothing saved, Settings → Memory says Shared facts the assistant learns for the whole team. Nothing saved yet. The empty list says Nothing yet. Patterns and facts from team conversations will show up here. With notes, it says {count} shared memories for the team. Updated {date}. AI memory says What the assistant keeps in mind to work the way you do. Empty: Nothing yet. As you work with the assistant it will note what matters — your routines, how you like replies written, facts it keeps needing — and those notes show up here for you to read, edit, or remove. The composer empty line is Nothing saved yet. Tell Rivvi something below. A search with no hits says No memories match that search. Memory is not Knowledge (the clinic and provider map) and not a file in the data lake.

​
Add, edit, remove

Both pages have Add memory, Edit memory, Remove memory, Version history, and Clear all. Save keeps a note. The toast is Memory added, Memory updated, or Memory removed. A missing name says Give it a short name. Empty text says Write what you want remembered. Settings → Memory opens Add an org memory. Seed a fact the assistant should know for everyone: legal name nuances, acronyms, specific rules. Don’t paste patient-specific info here. Name placeholder is operations/escalation-rules. A short label like operations/escalation-rules. Use a / to file it under a group. What to remember placeholder is Class-level facts about the org. Edit memory says The previous version is kept in history before this saves. The field is Content. Remove this memory? says This removes the current note. Earlier versions stay in history. Version history opens Earlier versions. With none: No earlier versions yet. This memory hasn’t changed since it was created. AI memory opens Add a memory. Tell the assistant something to keep in mind — a preference, a routine, a fact it keeps needing. Don’t paste patient-specific info here. Name placeholder is preferences/tone. A short label like preferences/tone. Use a / to file it under a group. What to remember says One to three short sentences is plenty. Edit memory says Earlier wording stays in history when you save. A machine-written note says This started as machine data. You’re editing the readable note; saving keeps your words. Remove this memory? says The assistant will stop using this note. You can add it again later if you change your mind. The confirm button is Remove. Version history opens Earlier versions.

​
Types

They are not the same list. Organization You

​
Clear all

Clear all on Settings → Memory asks Clear everything the assistant remembers for this org? This removes all {count} org memories. The assistant re-learns from future conversations. People’s personal memories and version history are not affected. One note says This removes the 1 org memory. The toast is Cleared {count} memories. Clear all on AI memory asks Clear everything the assistant remembers? This removes all {count} memories — your routines, preferences, and corrections. The assistant re-learns from future conversations. Org-shared memories and version history are not affected. The composer confirm is Remove every memory?

​
Say it in chat

You can still say “Remember that we don’t call on Fridays.” That writes a memory. Settings is where you review it, correct it, or remove it. On AI memory, Import opens Import memory. Paste notes from another assistant. Each heading becomes its own memory. The box is Paste memory here. One note saves as Add to memory. Several save as Add {count} memories. The box at the bottom is Tell Rivvi something to remember. On an open note it is Tell Rivvi what to change or remove. Save this note sends it. Delete and Keep are on that note. A row can say Summary, Show more, or Updated {date}. An empty note says No readable note for this entry yet.

​
Next

Your account

Knowledge