Memory Policy
Admin → AI & Tools → Memory Policy
The tenant-wide opt-out for AI memory. NudgeBee's assistant can carry context between conversations — your preferences, decisions you made during past investigations, patterns it has observed. This page decides which of those layers it may use for everyone in the tenant.
A change here affects the whole tenant. Individual users can restrict further in their own Privacy settings, but they cannot re-enable a layer you have turned off.
The master switch
Use memory in chats. When off, no layer is injected — nothing NudgeBee has remembered reaches a conversation. Every per-layer toggle is disabled while it is off.
The layers
| Layer | What it holds |
|---|---|
| Soul | Communication style, voice and values. |
| Preferences | Explicit defaults — namespace, cloud, channels. |
| Patterns | Behaviour patterns inferred across chats. |
| Decisions | Choices made during investigations. |
| Sessions | Short-term working memory within a chat. |
| Collective | Team-wide knowledge shared across users. |
Turn a layer off and it stops being injected into conversations. Each toggle records who changed it and when, so the tenant policy has an audit trail.
Turning a layer off does not delete it
This is the distinction that matters for a data-handling question: disabling a layer stops injection, it does not remove stored data. To delete what has been remembered, use the per-row delete actions in the memory views.
If you need both — stop using it and remove it — do both, in that order.
Choosing a policy
- Collective is the one to think hardest about: it shares knowledge across users, which is the point, and also means one user's context can surface in another's conversation.
- Sessions is short-term and scoped to a single chat, so it is the least sensitive.
- Decisions and Patterns are what make the assistant improve over repeated use. Turning them off costs you that.
Related
- Egress Filter — what is allowed to leave for the model
- Security