Guides

Set up semantic search with document ingestion and vector-based retrieval - no graph required.

Build a full graph RAG system with automatic entity extraction and graph-augmented retrieval.

Add persistent memory to your AI agents - store, recall, correct, and build prompts from thread history.

Use tenant boundaries, graph overlays, and bucket write routing to isolate public, internal, and private knowledge.

Configure and initialize TypeGraph in self-hosted mode - database adapters, embeddings, LLMs, and the deploy/init lifecycle.

Master typegraph.search() - resources, weights, context, graph retrieval, prompt building, and result ranking.

Inspect graph entities, facts, traces, traversal behavior, and when to use search graph weighting vs graph.explore() as a probing tool.

Understand TypeGraph document and event entity extraction from ingest-time LLM calls through entity resolution, fact records, and chunk links.

Learn how context, graph overlays, graph entities, memory subjects, and deterministic external IDs fit together.

Run TypeGraph on your own infrastructure - tables, cloud vs self-hosted differences, batch ingest, and schema management.

Build a production self-hosted multi-tenant RAG app with schema-per-tenant isolation, SDK factory pattern, and identity scoping.

Tightly scope PostgreSQL credentials for TypeGraph with least-privilege roles, schema permissions, and read-only access.

Instrument your pipeline with telemetry, trace threads end-to-end, monitor search performance, and export events.