Does Glen replace my data warehouse?

No. The warehouse holds your data; Glen holds the durable understanding of it, the conventions, caveats, and decisions your agents otherwise re-derive every session. They're complementary.

What knowledge should agents write to Glen?

Durable facts about your platform: schema conventions, deprecated tables, why a metric is defined a certain way, root causes of past incidents, anything an agent would otherwise rediscover next time.

Shared memory for data engineers

As a data engineer you own the pipelines, warehouses, and models that the rest of the company queries, and you're increasingly pointing AI agents at that surface to debug pipelines, document tables, and answer questions about the data. But those agents are amnesiac. Each session re-discovers your schema conventions, re-learns which tables are deprecated, and re-derives the tribal knowledge about why a column means what it means, knowledge that lives in your head and in scattered docs, not anywhere an agent can reach. Glen, shared memory for AI agents, gives every agent touching your data platform one durable, org-shared memory through a single MCP tool, so hard-won context about your data stops evaporating each session.

Connect Glen as an MCP tool and any agent working on your data stack can retrieve relevant context and record what it learns in one call. An agent debugging a broken DAG pulls what's already known about that pipeline's quirks and prior failures before diagnosing; an agent documenting a table reads the conventions and caveats your team established; an agent that figures out why a metric drifted writes the explanation back so the next investigation starts there. You stop re-explaining your warehouse to every fresh session and let the platform's institutional knowledge accumulate.

Because Glen is org-scoped, that knowledge belongs to the whole data function and beyond. One engineer's agent learns a durable fact about a dataset, the gotcha in a join, the reason a table is the source of truth, and every other agent reads it next time, no rediscovery, no knowledge stranded in a Slack thread. And because Glen is a standard MCP server, the same memory is reachable from Claude Code, your warehouse-querying agents, and your pipeline tooling, so the context spans the tools you actually use. You wire it in once over OAuth or an API key and let the data knowledge compound. For a data engineer, Glen complements the warehouse: the warehouse holds the data, Glen holds the durable understanding of that data your agents keep rebuilding.

FAQ

Does Glen replace my data warehouse?: No. The warehouse holds your data; Glen holds the durable understanding of it, the conventions, caveats, and decisions your agents otherwise re-derive every session. They're complementary.
What knowledge should agents write to Glen?: Durable facts about your platform: schema conventions, deprecated tables, why a metric is defined a certain way, root causes of past incidents, anything an agent would otherwise rediscover next time.