What is Chunking Strategy? The choice of how to split source documents into smaller pieces before embedding them for RAG retrieval. Affects retrieval accuracy more than any other RAG decision. Chunking Strategy explained Common strategies: fixed-size chunks (e.g. 512 tokens with 50-token overlap), semantic chunks (split at section or paragraph boundaries), agentic chunks (LLM decides splits), and hierarchical chunks (small + parent-doc lookup). Wrong chunking is the top cause of RAG returns garbage problems: too small loses context, too large dilutes the embedding. How Groovy Web uses Chunking Strategy We tune chunking strategy per client corpus. Legal docs need section-boundary chunks, code repos need function-level, and product docs work with semantic + parent-doc. Explore our RAG System Development service → Chunking Strategy: common questions What is Chunking Strategy? The choice of how to split source documents into smaller pieces before embedding them for RAG retrieval. Affects retrieval accuracy more than any other RAG decision. What should I know about Chunking Strategy? Common strategies: fixed-size chunks (e.g. 512 tokens with 50-token overlap), semantic chunks (split at section or paragraph boundaries), agentic chunks (LLM decides splits), and hierarchical chunks (small + parent-doc lookup). How does Groovy Web use Chunking Strategy? We tune chunking strategy per client corpus. Legal docs need section-boundary chunks, code repos need function-level, and product docs work with semantic + parent-doc. Which terms are related to Chunking Strategy? Chunking Strategy is closely related to Retrieval-Augmented Generation (RAG), Embeddings (AI), Vector Database and pgvector. Related terms Retrieval-Augmented Generation (RAG) Retrieval-Augmented Generation (RAG) is an AI architecture that combines a language model with a search system to answer questions using your own data. Embeddings (AI) Embeddings are numerical representations of text, images, or other data that capture semantic meaning, enabling AI systems to understand similarity and relationships. Vector Database A vector database is a specialized database designed to store, index, and search high-dimensional vector embeddings for AI applications like semantic search and RAG systems. pgvector pgvector is a PostgreSQL extension that adds vector similarity search capabilities, enabling RAG systems and semantic search directly in your existing PostgreSQL database. HNSW Algorithm Hierarchical Navigable Small World: the leading approximate-nearest-neighbor index used inside almost every modern vector database for fast similarity search at scale. Hybrid Search A search technique combining keyword matching with semantic similarity, leveraging both approaches to provide more relevant and comprehensive results than either alone. ← More in Retrieval & RAG · All 107 glossary terms