Real-world knowledge evolves. Temporal awareness allows GraphRAG to access and reason over the history of your data, understanding how patterns emerged, when relationships changed, and the context of events. This leads to far more accurate, relevant, and insightful answers than static RAG.

It means our LLM can actively access and reason over the symbolic structure of your temporal graph—its entities, relationships, and their changes over time—alongside leveraging semantic vector search. This gives it unparalleled freedom to explore and understand your data holistically, instead of relying on pre-compiled queries that can fail.

Fully customisable. You can define your own graph data vectorization using JINJA templates, choose any LLM (OpenAI, Gemini, open-source, local models), and build custom agentic workflows with multiple RAG pipelines using Python or Rust, all on top of a LlamaIndex foundation.

You can define multiple specialised RAG pipelines (e.g., for data lookup, semantic search, or running graph algorithms). An orchestrator AI then intelligently selects and combines these agents in real-time to best answer a complex query or automate a multi-step task.

Beyond just providing an answer, our system references the specific data points, graph patterns, trends, or risks it used from the temporal graph. This provides clear data provenance, essential for user trust, explainability, and meeting compliance requirements.

Yes. By combining the LLM's reasoning with direct access to real-time temporal graph analytics and the ability to trigger complex algorithms, Pometry GraphRAG can automate sophisticated data analysis and reporting workflows that would otherwise require significant manual effort and time.