What if a system has no connector?
Send its documents, messages, and JSON through the Zep SDKs.

Give agents one governed company brain, with every fact linked to its source.
One query searches documents, conversations, and data together.
Each fact traces to the message or document it came from.
When a decision changes, Zep marks the old fact invalid and keeps its dates.
Connectors sync company knowledge from the tools your team uses.
Set what each user and agent can retrieve. Retention rules and logs live in the same system as the knowledge.
Policies decide which shared context each user and agent can retrieve.
Set how long context is kept, with legal hold when required.
Logs record each request and whether policy allowed it.
The Context Lake unifies business data into governed context for agents.
Pick the channels, folders, and libraries to bring in.
Each fact keeps its dates and its source.
Agents query the graph through the Zep SDKs or MCP, filtered by policy.
Who owns the Atlas migration?
Project: Atlas migration
Status: paused 2026-10-02
Owner: Data platform team
Source: #atlas-eng“We can easily see Zep becoming a de facto partner in this layer of the enterprise agent stack.”
Send its documents, messages, and JSON through the Zep SDKs.
No. Zep doesn't enforce Slack or Google Drive sharing settings when agents retrieve context. You set access with Zep policies for each user and agent.
Yes. Through the Context MCP Server, agents can write to the graphs their connection allows.
Search returns documents. Zep turns documents and conversations into facts with sources and dates, so agents get current facts and the history behind them.
Start with one internal workflow. We'll help map its sources and access policies.