How is Glen different from Griptape's conversation memory and task memory?

Griptape's memory primitives keep a single structure's run coherent and are scoped to that session or pipeline. Glen is durable, org-shared long-term memory that outlives any structure and is read and written by every agent in your organization.

How do my Griptape structures connect to Glen?

Glen is a standard MCP server. Expose it as an MCP tool inside your structure, then call it from any task to retrieve context and record observations in a single round trip.

Shared memory for Griptape

Griptape gives you a structured Python framework for building agents, pipelines, and workflows out of composable tasks, drivers, and rulesets. It ships memory primitives, conversation memory and task memory, that keep state coherent inside a single structure or session, plus drivers that let you swap out the underlying store. What it does not give you out of the box is durable knowledge that survives across structures, across deployments, and across every other agent your organization runs. Glen, shared memory for AI agents, fills exactly that gap: a single MCP tool that lets your Griptape structures read long-term organizational context and write new facts back in one round trip, so the knowledge compounds instead of resetting every time a structure finishes.

Wire Glen into your Griptape structures as an MCP tool and any task in a pipeline or workflow can call one tool that both retrieves relevant long-term context and records what the run learned. An early task can pull what the organization already knows about a customer, an account, or a system before the structure decides how to proceed; a later task can write back the new facts it discovered. You stop standing up a separate vector store, embedding driver, and write path just to give Griptape durable cross-session memory, and you stop maintaining per-structure silos that forget everything the moment they complete.

Griptape's conversation memory and task memory are scoped to a structure's lifetime, they keep a run coherent, but they are not a shared, durable knowledge base. Glen is org-scoped, so the memory layer spans every Griptape structure, every deployment, and every other agent in your organization rather than being bound to one Python process. One pipeline learns a durable fact; a different workflow, even one written in another framework entirely, reads it the next time it runs. Because Glen is a standard MCP server, the same memory your Griptape agents write is also readable from Claude Code, Cursor, or any other MCP client, so human-driven and structure-driven work draw on one shared store. You connect it once over OAuth or an API key and let the organizational knowledge accumulate across every run.

FAQ

How is Glen different from Griptape's conversation memory and task memory?: Griptape's memory primitives keep a single structure's run coherent and are scoped to that session or pipeline. Glen is durable, org-shared long-term memory that outlives any structure and is read and written by every agent in your organization.
How do my Griptape structures connect to Glen?: Glen is a standard MCP server. Expose it as an MCP tool inside your structure, then call it from any task to retrieve context and record observations in a single round trip.