Knowledge & Training

Building Your Knowledge Base

Your agent's knowledge base is the foundation of its intelligence. The better your knowledge base, the more accurate and helpful your agent's responses will be.

Fetchply supports four types of knowledge sources:

How It Works

When you add a knowledge source, Fetchply processes it through a training pipeline:

1. Content extraction — text is pulled from your source (web pages, files, etc.)
2. Chunking — content is split into segments of up to 512 tokens
3. Embedding — each chunk is converted into a vector using Voyage AI
4. Indexing — vectors are stored in the knowledge base for fast retrieval

When a visitor asks a question, Fetchply searches the knowledge base for the most relevant chunks and uses them to generate an accurate answer.

Storage Limits