Working with Documents
Leverage your data to make your agents smarter
The Agent SDK makes it trivial to work with Retrieval Augmented Generation (RAG) on your content. In this section, we'll explore connecting to different embedding stores, how to create custom content processors, and controlling where you store files as they're uploaded and processed.
At the heart of RAG are two interfaces, Document and Metadata. The Document holds the content section that you are embedding and the metadata provides additional context about that section. Metadata could be things such as the slide number of a presentation or how many images are on that slide.
This section from Langchain4J describes these fairly well.
Last updated
Was this helpful?