How is Glen different from Pydantic AI's run context and dependencies?

The run context and dependencies carry typed state for the duration of one run. Glen is durable, org-shared long-term memory that outlives any run and is read and written by every agent in your organization.

How do my Pydantic AI agents reach Glen?

Glen is a standard MCP server, and Pydantic AI supports MCP servers natively. Attach it to an agent and call it during a run to retrieve context and record observations in one round trip.

Shared memory for Pydantic AI

Pydantic AI brings type-safe, validated structure to agent building: typed dependencies, structured outputs, and a clean way to pass state through a run via the run context and dependency injection. That model is great for correctness within a single agent run, you know exactly what shape your data is. But the run context is ephemeral; it exists for the duration of one `run` call and then it is gone. The durable knowledge your organization's agents should carry between runs and share with each other is not what a dependencies object is for. Glen, shared memory for AI agents, supplies that long-term, org-wide memory as a single MCP tool every Pydantic AI agent reads from and writes to.

Add Glen to your Pydantic AI agents as an MCP server, which Pydantic AI supports natively, and each agent gains one tool that retrieves relevant long-term context and records new facts in a single round trip. At the start of a run, the agent can pull what the organization already knows so its typed reasoning is grounded in real history; when the run produces durable facts, it writes them back, all without you bolting a separate database and retrieval layer onto your otherwise clean, typed agent.

Pydantic AI gives you confidence in the shape of data within a run; Glen gives you continuity of knowledge across runs and across agents. Because the memory is org-scoped rather than tied to one run's dependencies, a fact one agent records is available to every other Pydantic AI agent in your org, and to agents built on other frameworks, because Glen is a standard MCP server, not a Python-only construct. The knowledge compounds run over run instead of resetting each time the context is rebuilt. Connect once over OAuth or an API key, and the same store your Pydantic AI agents write is readable from Claude Code, Cursor, or any other MCP client, so structured automated work and human-driven work share one source of truth.

FAQ

How is Glen different from Pydantic AI's run context and dependencies?: The run context and dependencies carry typed state for the duration of one run. Glen is durable, org-shared long-term memory that outlives any run and is read and written by every agent in your organization.
How do my Pydantic AI agents reach Glen?: Glen is a standard MCP server, and Pydantic AI supports MCP servers natively. Attach it to an agent and call it during a run to retrieve context and record observations in one round trip.