Zep

About

Zep builds temporal knowledge graphs from chats and business data, then returns a context block for your agent's prompt. It is a hosted service with a free monthly credit allowance; its graph engine, Graphiti, is open source.

What you can do with it

• Give a chat agent memory of each user's facts and preferences
• Merge business data and conversations into one knowledge graph
• Retrieve a ready-made context block for the next LLM prompt

Get started

1. Create a project and API key in the Zep dashboard
2. Install the SDK with pip install zep-cloud

Example

import os
from zep_cloud import Zep, Message

client = Zep(api_key=os.environ["ZEP_API_KEY"])
client.user.add(user_id="user-123", first_name="Jane")
client.thread.create(thread_id="thread-123", user_id="user-123")
client.thread.add_messages("thread-123", messages=[Message(role="user", content="I'm vegetarian.")])
print(client.thread.get_user_context(thread_id="thread-123").context)

Details

Hosting

Hosted service

Available in

Worldwide

Official SDKs

Python, JavaScript/TypeScript, Go

MCP server

Remote