Skip to main content

​
Understanding Chunking

When you ingest a large document or a massive webpage, ZynfoAI doesn’t just read it as one massive block of text. Instead, it systematically breaks the text down into smaller, highly optimized pieces called Chunks.

​
Why Chunking Matters for AI Agents

Chunking is the foundation of Retrieval-Augmented Generation (RAG). It allows the AI agent to perform highly accurate Vector Searches. When a user asks a question, the AI agent mathematically searches through all your chunks to find the 3-5 chunks that are most contextually similar to the user’s question. It then reads only those specific chunks to formulate its exact answer.

​
Viewing and Auditing Chunks

You can transparently inspect exactly how ZynfoAI has broken down your documents:
  1. Go to the Knowledge page.
  2. Find a specific source and click the three-dot menu (More Options).
  3. Select View Chunks.
You will be taken to the /knowledge/chunks dashboard where you can read every individual paragraph or sentence segment that the AI agent has memorized from that specific source. This transparency is incredibly helpful for debugging why an AI agent might be giving a slightly inaccurate answer.