Context, Memory, and Retrieval
Agentic RAG
Ordinary RAG retrieves a fixed set of documents once and answers from them. Agentic RAG adds agents that decide what to retrieve, retrieve repeatedly, judge the quality of what came back, and adapt.
4 chapters31 slides19 minUpdated
Your route through the subject
0/31 slides completed · 0/4 chapters · 0 slides read
Start with: From Naive RAG to Agentic RAG: The Evolution
The chapter map
Follow the route or choose a chapter.Chapters
Chapter 1: From Naive RAG to Agentic RAG: The Evolution
Traditional RAG retrieves a fixed set of documents once and answers from them. Agentic RAG evolved through five stages into agents that control retrieval.
4 min5 slides: From Naive RAG to Agentic RAG: The Evolution0/5 complete
Chapter 2: The Building Blocks of Agentic Intelligence
An AI agent combines a model, memory, planning, and tools. Four agentic patterns and five workflow patterns are the building blocks agentic RAG combines.
5 min6 slides: The Building Blocks of Agentic Intelligence0/6 complete
Chapter 3: The Taxonomy of Agentic RAG Systems
Seven agentic RAG architectures in order of rising autonomy, from a single-agent router through hierarchical, corrective, adaptive, and graph-based designs.
6 min10 slides: The Taxonomy of Agentic RAG Systems0/10 complete
Chapter 4: Practical Guidance: When and How to Use Agentic RAG
Agentic RAG is not a universal default. Retrieval quality is usually the real bottleneck, and added autonomy needs explicit constraints and stopping criteria.
4 min10 slides: Practical Guidance: When and How to Use Agentic RAG0/10 complete