Does Glen replace my Haystack document store?

No. Your document store grounds answers in indexed content. Glen adds durable, org-shared memory of what agents learn while working, which complements retrieval rather than replacing it.

How do my Haystack pipelines reach Glen?

Glen is a standard MCP server. Connect to it as an MCP tool and call it from any pipeline component or agent to read context and write observations in one round trip.

Shared memory for Haystack

Haystack is deepset's framework for building search and retrieval-augmented pipelines, document stores, retrievers, rankers, and generators composed into production RAG and agent flows. It is excellent at grounding answers in your indexed documents, but document retrieval is not the same as memory: the durable facts your pipelines discover during a run, the corrections a user makes, the state of an ongoing task, none of that flows back into your index automatically or gets shared across pipelines. Glen, shared memory for AI agents, gives your Haystack pipelines long-term shared memory as a single MCP tool, complementing your RAG index with knowledge that agents write as well as read.

Expose Glen to a Haystack pipeline over MCP and any component or agent can call one tool that retrieves relevant long-term context and records new facts in a single round trip. Your retrievers still ground answers in your document store; Glen captures the things documents do not, the facts an agent establishes, the preferences a user states, the outcomes of prior runs, so the next pipeline does not start from zero.

The distinction matters. A Haystack document store holds the corpus you indexed; Glen holds what your agents learn while working. And where a document store is typically scoped to one pipeline or app, Glen is org-scoped: the memory is shared across every Haystack pipeline, every agent, and every deployment in your organization. One pipeline learns a durable fact; another reads it next time, even a pipeline running a different model or built on a different framework. Because Glen is a standard MCP server, the same memory is readable from Claude Code, Cursor, or any other MCP client, so human-driven and pipeline-driven work draw on one shared knowledge layer. You connect once over OAuth or an API key and let retrieval and memory reinforce each other.

FAQ

Does Glen replace my Haystack document store?: No. Your document store grounds answers in indexed content. Glen adds durable, org-shared memory of what agents learn while working, which complements retrieval rather than replacing it.
How do my Haystack pipelines reach Glen?: Glen is a standard MCP server. Connect to it as an MCP tool and call it from any pipeline component or agent to read context and write observations in one round trip.