Grounded AI learning hub
RAG and Knowledge Systems
RAG and knowledge systems give AI applications access to current, attributable evidence. A reliable system must preserve source identity, enforce permissions before information can influence the model, evaluate retrieval separately from generation, and know when graph structure adds value beyond text similarity.
Best for: Developers building grounded assistants, search systems, and knowledge-aware agents
Recommended learning order
- 1Embeddings & Hybrid SearchTokenization, dense and sparse representations, similarity, ANN indexes, quantization, and rank fusion.Open lesson
- 2RAGRetrieval Augmented Generation for grounding AI.Open lesson
- 3Knowledge Graphs & OntologiesRDF triples, Turtle, OWL, SPARQL, SHACL, taxonomies, schemas, and semantic reasoning.Open lesson
- 4Production RAG & GraphRAGChunking, query transformation, hybrid retrieval, reranking, entity resolution, and global graph search.Open lesson
- 5Knowledge-Driven Agent SystemsOntology-constrained plans, graph-augmented reasoning, live knowledge updates, and enterprise architecture.Open lesson
- 6LLM Interfaces & Structured OutputsContext construction, prompt contracts, tool calls, schema validation, retries, and observable model APIs.Open lesson
Decisions this path should help you make
- When should retrieval be lexical, dense, hybrid, or graph-based?
- How should permissions constrain ranking and context construction?
- Which metric diagnoses retrieval versus generation failure?
- When does an ontology improve decisions rather than add ceremony?