Knowledge Graph
A structured map of concepts and their relationships — the Elvish taxonomy of everything your AI needs to understand.
A knowledge graph represents information as a network of entities and relationships rather than as unstructured text. Nodes in the graph represent entities — a customer, a product, a company, a person, a concept — and edges represent the relationships between them: "Acme Corp uses Product X," "Product X competes with Product Y," "Decision Maker Z reports to Executive W." The structured, relational nature of knowledge graphs makes them distinct from document-based retrieval systems: while vector databases excel at finding semantically similar text, knowledge graphs excel at answering relational queries — "Who are all the decision-makers at accounts that use Feature A and have shown churn risk signals?" — that would require complex inference from unstructured documents.
Knowledge graphs and vector databases serve complementary purposes in AI systems. Vector databases store and retrieve unstructured knowledge efficiently based on semantic similarity. Knowledge graphs store and traverse explicitly structured relational knowledge. Many production AI systems use both: a vector database for retrieving relevant documents and a knowledge graph for answering questions about relationships, hierarchies, and structured facts. The combination allows an AI system to find relevant context through semantic search and reason about explicit relationships through graph traversal — producing answers grounded in both unstructured knowledge from documents and structured facts about how entities in the system relate to each other.
For B2B teams, knowledge graphs are most valuable when relational knowledge about customers, products, and markets is central to the AI system's usefulness. A sales intelligence AI that needs to know which accounts are connected, who influences whom within buying committees, and how competitive products relate to your own requires structured relational knowledge that document retrieval alone can't efficiently provide. A product recommendation AI that needs to reason about feature relationships, user segments, and usage patterns benefits from explicit graph structure. Building and maintaining a knowledge graph requires more upfront data modeling work than a simple document store, but pays off when the questions the AI must answer are fundamentally about relationships rather than content.
Related terms
- Retrieval-Augmented Generation (RAG)— Giving your LLM access to the Restricted Section — so it answers from real knowledge instead of confident hallucination.
- Embedding— Numeric representation of meaning — Elvish rune encoding, but as floating-point vectors optimized for semantic search.
- Vector Database— The database that stores meaning as numbers — the Sorting Hat's filing system, indexed for semantic retrieval.
- Grounding— Anchoring AI responses in verifiable data — insisting the model look at the actual Palantír rather than guess what it shows.
- Semantic Search— Finding results by meaning rather than keyword match — 'Accio relevant context' without knowing the exact incantation.