Knowledge Base
A knowledge base is the layer between your organisation's information and Claude's ability to use it. This section covers how to design, build, and maintain knowledge systems — from simple document stores to production-grade retrieval infrastructure.
In This Section
Knowledge Base Concepts
Documents vs structured data vs live APIs as knowledge — what kinds of knowledge Claude can access and how.
Document Ingestion
Loaders, parsers, and format handling — processing PDF, HTML, DOCX, and CSV into indexable content.
Chunking Strategies
Fixed-size, semantic, and hierarchical chunking — what works for different document types and query patterns.
Metadata & Filtering
Adding metadata to chunks for scoped retrieval and filtered search — schema design and query patterns.
Claude Projects vs Custom RAG
When Claude Projects is enough vs when you need your own retrieval stack — a practical decision guide.
Knowledge Base Use Cases
Customer support bot, internal wiki, codebase Q&A, research assistant — real-world patterns and trade-offs.