Manage your RAG knowledge base (corpus)

This document describes how to manage your RAG knowledge base (corpus). This document covers the following topics:

Corpus management

A corpus is a collection of documents that you can query to retrieve relevant contexts for response generation. When you create a corpus for the first time, the process might take an additional minute.

You can perform the following operations on a corpus:

Operation Description Parameters Examples
Create a RAG corpus Create a RAG corpus to import or upload documents.

If you're using VPC-SC and it's the first time that you've called `createRagCorpus` and the call returns an error, then just wait a few minutes and retry.

Create parameters v1beta1
Create parameters v1
Create example v1beta1
Create example v1
Update a RAG corpus Update a previously-created RAG corpus to import or upload documents. Update parameters v1beta1
Update parameters v1
Update example v1beta1
Update example v1
List RAG corpora List all of the RAG corpora. List parameters v1beta1
List parameters v1
List example v1beta1
List example v1
Get a RAG corpus Get the metadata describing the RAG corpus. Get parameters v1beta1
Get parameters v1
Get example v1beta1
Get example v1
Delete a RAG corpus Delete the RAG corpus. Delete parameters v1beta1
Delete parameters v1
Delete example v1beta1
Delete example v1

Concurrent operations on corpora aren't supported. For more information, see the RAG API reference for v1beta1 or RAG API reference for v1.

File management

You can add files to your corpus by either uploading them directly or importing them from another storage location. The following table compares these two methods.

Method Description Use Case
Upload a RAG file Uploads a single file from your local machine to the RAG corpus. Best for adding individual files or a small number of files quickly.
Import RAG files Imports a batch of files from a storage location like Google Cloud Storage. Ideal for adding a large number of files at once or for automated workflows where files are already in cloud storage.

You can perform the following operations on files in a corpus:

Operation Description Parameters Examples
Upload a RAG file Upload a file from local storage with additional information that provides context to the LLM to generate more accurate responses. Upload parameters v1beta1
Upload parameters v1
Upload example v1beta1
Upload example v1
Import RAG files Import a set of files from some other storage into a storage location. Import parameters v1beta1
Import parameters v1
Import example v1beta1
Import example v1
List RAG files List all of the files in the RAG corpus. List parameters v1beta1
List parameters v1
List example v1beta1
List example v1
Get a RAG file Get details about a RAG file for use by the LLM. Get parameters v1beta1
Get parameters v1
Get example v1beta1
Get example v1
Delete a RAG file Delete a file from the RAG corpus. Delete parameters v1beta1
Delete parameters v1
Delete example v1beta1
Delete example v1

What's next