Contexts (Memory)

Contexts give your agents memory. They store structured information that the agent can use across conversations and runs. This makes agents more consistent, personal, and useful.

How Contexts Work

  • Context Buckets define what kind of memory an agent has.
    Example: User Profile, Project Settings, Session Notes

  • Contexts are the actual pieces of memory stored in those buckets.
    Example: User Profile might store { name: "Asha", tone: "concise" }.

Whenever you run an agent:

  • If you provide a context, the agent will use it.
  • If you don't, Fluo will create a default one automatically.
  • The agent can also update contexts as conversations evolve.

Why Use Contexts?

  • Personalization: Remember user details or preferences.
  • Continuity: Keep track of ongoing projects or sessions.
  • Collaboration: Share project state across multiple agents in the same project.

Examples

  • User Profile: Store a customer's name, email, preferences.
  • Project State: Track the current phase or deadline of a project.
  • Whiteboard Context: Hold notes and ideas from a collaborative workspace.

How to Configure Contexts

  1. Create a Context Bucket
    Decide what information you want the agent to remember (e.g. user_profile).

  2. Define Fields
    Specify the fields (e.g. name, tone, last_active).

  3. Attach to an Agent
    Choose which agents should have access to this memory.

  4. Use in Conversations
    When you run the agent, you can provide context values, or let Fluo auto-fill them.

Things to Know

  • Contexts are small and structured — use them for state, not for long documents.
  • Each context belongs to a project, and can optionally be tied to a specific agent or end-user.
  • You can update contexts as the agent learns new information.

👉 Think of Contexts as smart sticky notes for your agent. They give it just enough memory to act consistently and stay relevant, without overwhelming it.