Builds and queries temporally-aware knowledge graphs tailored for AI agents operating in dynamic environments.
Graphiti is a framework designed to create and query dynamic knowledge graphs, specifically for AI agents. It facilitates the integration of user interactions, structured data, and external information into a coherent, queryable graph. Unlike traditional RAG methods, Graphiti supports real-time incremental updates and historical queries without recomputation, making it ideal for interactive, context-aware AI applications.
Key Features
01Real-time incremental updates for dynamic data integration
02Bi-temporal data model for accurate historical queries
037,402 GitHub stars
04Hybrid retrieval combining semantic, keyword, and graph-based search
05Scalable architecture for large datasets with parallel processing
06Custom entity definitions with Pydantic models
Use Cases
01Integrate and maintain dynamic user interactions and business data
02Enable state-based reasoning and task automation for agents
03Query complex, evolving data with semantic, keyword, and graph-based search
