Twin Memory

Your Digital Twin is the assistant you talk to. Twin Memory is what it remembers between those conversations, so you don’t have to repeat yourself. Twin Memory has two parts:

---

​

Long-term Memory — what your twin keeps on its own

Most assistants forget everything the moment a conversation ends. With Long-term Memory switched on, yours doesn’t: when something in a conversation genuinely changes how it should help you later, it keeps that. You don’t have to ask. There’s no “save this” button to remember to press. If you do ask — “remember that I prefer metric units” — it saves that too, and confirms it. Most exchanges leave nothing behind. It keeps what matters, not a transcript. It works the same when you talk to your twin in Google Chat: what matters is kept, and the memories show on that conversation when you open it on the website. In a shared space, nothing is kept unless you confirmed sharing for that message.

​

You always see what it remembered and what it saved

Whenever an exchange touches your twin’s long-term memory, a single Memory row appears under the answer, in the same expandable style as the Search Results row. Open it to see every memory involved, each tagged with what happened to it:

• Recalled — a memory your twin brought back to shape this answer, with how it was found.
• Saved — something new it kept from this exchange.
• Updated — an existing memory it revised.
• Forgotten — a memory it removed.

Click a memory to read exactly what was kept. If you ask for something to be kept confidential — or your twin judges it so on its own — the row lists the memory without showing the content, and you have to click Reveal to see it. Saying so yourself always counts: if you call something confidential, private, or secret, your twin treats it that way regardless of how it would have classified it. It will use the memory when your question genuinely calls for it, and won’t bring it up unprompted in unrelated conversation. A memory you have since deleted says so rather than showing stale content. Your administrator can hide the Recalled memories for a quieter transcript (Display Memory Details); memories keep working either way, and anything your twin saves, updates or forgets always stays visible in the row.

​

Managing what it remembers

Everything lives in Settings → Memory (new interface) or Manage Memory… → Long-term Memory (classic interface, from the [+] action menu’s Knowledge submenu — not sure which interface you have? See Classic interface / New interface in the glossary), where you can:

• Browse and search everything your twin has stored
• Correct anything it got wrong or worded badly
• Export the whole lot as a file, including anything confidential
• Delete any memory permanently

---

​

Shared Memory — what your organization shares

The second half of Twin Memory is the information everyone using the same Digital Twin can draw on: opening hours, policies, key contacts, anything the whole team benefits from. This works exactly as it always has — see Shared Memory for the full picture.

---

​

Availability

Long-term Memory is switched on for a Digital Twin by the Praxis AI team, not by the twin’s own administrators. They see the switch under Instance Settings → Knowledge → Long-term Memory, but it is greyed out for them, with a line saying so. To have it enabled for a twin, ask Praxis AI. Once it is on, the twin’s own administrators decide how far it reaches: the default is every member, with Long-term Memory as the default mode, and they can narrow it to administrators only while they evaluate it. If you don’t see a Long-term Memory section under Settings → Memory (or a Long-term Memory tab in Manage Memory…), it isn’t enabled for you. It is available in both interfaces.

For administrators: the toggle inside a conversation

The Knowledge submenu of the classic [+] menu, and the Memory section of the new interface’s + menu, carry an Enable Long-term Memory toggle; the classic submenu’s summary line shows whether it is currently on. Digital Twin administrators see this toggle in both places and in Instance Settings, but it is greyed out — the Praxis AI team turns long-term memory on or off for a Digital Twin. To ask for it, contact humans@praxis-ai.com.What a Digital Twin’s own administrators can change, once it is on: Admins only, which keeps long-term memory to administrators while they evaluate it. Sharing memories across the Digital Twins of an account is also a Praxis AI setting.

Where it is available, you can still choose for yourself. Settings → Memory (or Manage Memory… → Long-term Memory) lets you pick whether your twin builds long-term memories on its own, or sticks to only the facts you save by hand.

---

​

How it compares

See How Memory Handles Your Data for what is stored, who can see it, and how to remove it.