Graphiti

What it does

Graphiti is an open-source framework, originated by Zep, that turns conversations, business data, and documents into temporal Context Graphs made of entities, relationships, and the timeline they live on. As facts change, outdated ones are invalidated and kept as history instead of being deleted. Retrieval combines vector similarity, full-text search, and graph traversal in one call.

Developers building AI agents can use Graphiti as the memory and context layer behind their applications. An MCP server connects it to MCP-compatible clients such as Claude and Cursor. Every edge carries timestamps, so point-in-time queries follow from the data model. It runs on Neo4j, FalkorDB, or Amazon Neptune and supports OpenAI, Azure OpenAI, Google Gemini, and Anthropic models.

Graphiti suits teams that want to run graph-backed memory themselves; the framework can be run locally. The graph database and LLM provider are configured separately and are not bundled with the open-source code. Zep's managed platform is a separate commercial offering and is not required to use Graphiti.

What it can do

• Builds temporal knowledge graphs from conversations, business data, and documents
• Invalidates outdated facts while preserving them as history
• Hybrid retrieval combining vector, full-text, and graph traversal in one call
• Point-in-time queries based on temporal edge timestamps
• Custom entity and edge types for domain-specific knowledge
• MCP server for connecting to Claude, Cursor, and other MCP clients

Works with

• Neo4j
• FalkorDB
• Amazon Neptune
• OpenAI
• Azure OpenAI
• Google Gemini
• Anthropic
• Model Context Protocol (MCP)

Before you choose

• Running Graphiti requires your own graph database and LLM provider, which may carry separate costs.
• The open-source license terms are not stated on the Graphiti pages; check the repository before adopting.
• Graphiti is distinct from Zep's commercial managed platform; confirm which product a page describes.

Listed under

• Coding
• MCP and agent tools
• Agent memory
• Developers
• Knowledge graph
• Mcp server
• Open source
• Retrieval

Get started

Quick Start covers installation, episodes, and search.

1. Read the Graphiti Quick Start for installation and basic operations.
2. Choose and configure a supported graph backend such as Neo4j, FalkorDB, or Amazon Neptune.
3. Configure an LLM provider from the supported list and add episodes through the documented APIs.

Graphiti is the open-source framework and is distinct from the managed Zep platform. No hosted subscription is implied; budget for your own LLM provider, graph database and hosting.

Graphiti is free to use. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Graphiti | Zep ↗www.getzep.com
• Welcome to Graphiti! | Zep Documentation ↗help.getzep.com
• Pricing | Zep ↗www.getzep.com

Company website

getzep.com: product, pricing and policy pages

Hacker News

Launch posts and discussion

YC directory mirror

Batch and company name

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

manual_review

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.