Data & Storage · Integration
Zep
Add Zep to your product for your customers, and give your AI agents governed access to it.
Zep is where an AI application keeps what it already knows about a person. Messages go into a thread against a user, and Zep folds them into a temporal knowledge graph whose nodes are entities and whose edges are the facts connecting them, then hands back a context block holding a summary of that user and the facts most relevant to the conversation in front of them. The temporal part is what makes it worth embedding rather than reimplementing: when something changes, the superseded fact is marked invalid from a date instead of being deleted, so an application can answer what is true now and still show what was true in March. Custom entity types let each customer shape what gets extracted, which turns the graph into a per-tenant schema question rather than a fixed one. Every customer connects their own project and key, so their memory sits under an account they own, and fastn keeps the credentials and the client current.
In your product
Embedded for your customers. Per-tenant auth, no per-customer code, maintained by fastn.
Let each customer connect their own Zep project so the memory your product accumulates about their users stays in an account they own and pay for.
Add messages to a thread as a conversation happens and let the graph build itself, rather than maintaining a summary table nobody trusts.
Retrieve the context block before a model call, so an answer is grounded in what your product learned months ago and not only in the last few turns.
Define custom entity types per customer, because what is worth extracting from a conversation differs between one deployment and the next.
For your AI agents
Governed, audited access for the agents you build, through the MCP server.
An agent recalls something a user said weeks ago and cites the fact along with the date it became true.
An agent writes a new fact into the graph within scoped permissions, attributed to the tenant that authorised it.
An agent notices a stored fact has been superseded and says what replaced it instead of repeating the stale one.
Example prompt
What do we already know about this user, and which of those facts are still valid today?
Set up Zep in 4 steps
- 01Enable the Zep connector in your fastn dashboard.
- 02Have each customer supply a key for their own Zep project, so their users' memory is held under their account.
- 03Decide which users and threads your product writes, declare any custom entity types you depend on, and choose where the context block is fetched.
- 04Call it from your product and expose the same connection to your agents through the MCP gateway.
Why teams use the Zep integration
What you get by embedding it with fastn instead of building it yourself.
- Ship a Zep integration without building it. Your customers connect their own Zep account inside your product and work their records, datasets and fields there, with no per-customer code on your side.
- Handle the part that actually costs time: schemas differ per customer and change without notice, and volumes can be large. fastn owns the auth, token refresh, rate limits, pagination and breaking-change fixes, so a Zep update is not your on-call problem.
- One integration serves your product and your agents. The same governed Zep connection powers in-product features and gives AI agents scoped, audited access, so you read and write your customers' data where it already lives without wiring it twice.
Used by these teams
Compare with
Pinecone
QdrantWorks well with
Anthropic ClaudeOften used alongside
Tools the same teams tend to run next to Zep, across other categories.
OpenAI
Azure OpenAI
Hugging FaceZep integration FAQ
How do I add a Zep integration to my product?
Enable the Zep connector in your fastn dashboard, then let each customer authenticate their own Zep account. fastn handles the OAuth flow, token storage and refresh per tenant, so there is no Zep client code in your app and no per-customer branch in your codebase. Setup is 4 steps.
Do my customers each connect their own Zep account?
Yes. Every connection is scoped to the individual customer, so each authorises their own Zep account and only ever sees their own records, datasets and fields. That per-tenant isolation is the point of an embedded integration: you support the long tail of customer setups without maintaining an integration per customer.
Can AI agents use this Zep integration?
Yes. The same connection is exposed to your agents through the fastn MCP gateway, with permissions scoped per tenant and every call audited. An agent recalls something a user said weeks ago and cites the fact along with the date it became true.
Who maintains the Zep integration?
fastn does. When Zep changes an endpoint, deprecates a field or alters its auth, the fix lands in the connector rather than in your backlog, and your customers' connections keep working.
Does the Zep integration adapt when a customer's schema changes?
Schema and field mapping is configuration per customer, so a change on their side is a mapping update rather than a code change and a release on yours.
How are large Zep reads handled?
Pagination and throttling are handled for you, and initial backfills are rate-limited so a large import does not exhaust a customer's API allowance.
What can I build with the Zep integration?
A common starting point: let each customer connect their own Zep project so the memory your product accumulates about their users stays in an account they own and pay for. Teams also use it for the other use cases listed above, and expose it to agents for governed reads and writes.
How much does the Zep integration cost?
It is included. Pricing is based on connected accounts, not on how many connectors you enable, so adding Zep does not change your per-connector cost. You can start free with 3 connected accounts.
Add Zep to your product
Start free with 3 connected accounts. No sales call required, and no per-customer integration code.