How is Glen different from CopilotKit's readable state and actions?

CopilotKit's state and actions make a copilot aware of the current app and session. Glen is durable, org-shared long-term memory that persists across sessions and users and is read and written by every agent in your organization.

How do I connect Glen to a CopilotKit copilot?

Glen is a standard MCP server. Register it as an MCP tool in the agent runtime behind your copilot, then call it during an interaction to retrieve context and record observations in one round trip.

Shared memory for CopilotKit

CopilotKit lets you drop AI copilots and in-app agents straight into your React application, wiring your UI state, actions, and backend into a chat or agentic experience your users drive in context. It gives you readable app state, frontend actions the copilot can invoke, and a clean bridge to agent runtimes, but it deliberately leaves long-term memory to you. Each copilot session knows what is on screen and what you feed it; it does not remember what a previous user did, what another copilot in your product learned, or what the rest of your organization's agents already know. Glen, shared memory for AI agents, gives CopilotKit copilots a durable, org-shared memory as a single MCP tool that reads context and writes new facts in one round trip.

Add Glen as an MCP tool behind your CopilotKit agent and any copilot interaction can retrieve relevant long-term context and record what it learned in a single call. When a user starts a task, the copilot pulls what the organization already knows, prior decisions, preferences, the state of an account, so it does not ask the user to repeat themselves; when the task finishes, it writes back the durable facts worth keeping. You skip building a separate memory backend, embedding pipeline, and per-user store just to make your in-app copilot feel like it remembers, and you avoid each copilot instance starting from a blank slate.

CopilotKit's readable state and actions are scoped to the current session and the current screen; they make the copilot aware of the app, not of history. Glen is org-scoped, so the memory spans every copilot in your product, every user session, and every other agent your organization runs, in any framework. One copilot learns a durable fact and the next session, or a backend agent, or a teammate's copilot, reads it. Because Glen is a standard MCP server, the same memory your CopilotKit copilots write is readable from Claude Code, Cursor, or any MCP client, so the knowledge your product's AI accumulates is not trapped in the browser. Connect once over OAuth or an API key and let the in-app memory compound across users and sessions.

FAQ

How is Glen different from CopilotKit's readable state and actions?: CopilotKit's state and actions make a copilot aware of the current app and session. Glen is durable, org-shared long-term memory that persists across sessions and users and is read and written by every agent in your organization.
How do I connect Glen to a CopilotKit copilot?: Glen is a standard MCP server. Register it as an MCP tool in the agent runtime behind your copilot, then call it during an interaction to retrieve context and record observations in one round trip.