Building Knowledge Systems for AI: Graphs, RAG, Memory, and Context
Published by Self-published · Tech
A deep dive into building robust knowledge systems for AI using knowledge graphs, RAG, memory architectures, and context management.
Who this book is for
AI engineers designing knowledge retrieval for agents, knowledge engineers and data architects building enterprise knowledge infrastructure, search engineers moving to AI-powered retrieval, and technical leaders evaluating knowledge system architectures.
Table of contents
- The Knowledge Problem in AI
- Knowledge Graphs: Foundations
- Ontologies and Schema Design
- Entity Linking and Resolution
- Embedding Models and Vector Search
- Hybrid Search Systems
- Building Knowledge Graphs from Unstructured Data
- GraphRAG: Knowledge Graphs Meet Retrieval
- Semantic Indexing and Structured Retrieval
- Agent Memory and Knowledge Integration
- Multimodal Knowledge Systems
- Governing Knowledge Systems
Excerpt
A product team at a mid-size financial services company built a customer support chatbot in three days. They connected GPT-5.4 to a simple chat interface, wrote a system prompt describing the company's products, and launched it internally. The first day went well. On the second day, a customer asked about a pricing change that had been announced the previous week. The chatbot confidently quoted the old pricing. On the third day, a customer asked whether their account qualified for a recently introduced loyalty discount. The chatbot invented a discount program that did not exist.
