Skip to main content
An Actor represents a participant identity in MemoryLake. Every message in a conversation is attributed to an actor, and every fact extracted from those conversations is associated with one. Actors are the subject of memory — they are what MemoryLake “remembers about.”

​
Actor Types

​
How Actors Work

  1. Create an actor globally with a custom_id that maps to your user ID
  2. Bind the actor to one or more workspaces
  3. Use the actor’s ID when appending messages to conversations
  4. Query the actor’s accumulated facts via Actor Facts

​
Actor Facts vs Project Facts

MemoryLake stores facts at two scopes:
  • Actor Facts — knowledge about a specific person or agent, accumulated across all projects in a workspace. Example: “Jane prefers dark mode”
  • Project Facts — knowledge scoped to a specific project. Example: “Q1 revenue was $12.5M”
When an actor participates in conversations across multiple projects, their actor facts follow them — enabling per-user memory across your entire application.

​
Endpoints