Create アマゾン ナレッジ ベース

This is a prerequisite step before you create a ナレッジ ベースに基づく モデル接続.

To use the Amazon Bedrock-RAG capability, you would first create a Knowledge Base in Amazon Bedrock.

The アマゾン ナレッジ ベース comprises of a data source, an embedding model, and a vector store that you would configure in the Amazon Bedrock console in the following order:
  1. Within Amazon Bedrock, navigate to Knowledge Bases to create a knowledge base.
    注: See Create a knowledge base .
  2. Next, you would choose your data source from options such as Amazon S3, web crawler, Confluence, Salesforce, or SharePoint, and configure it with the default settings.
    注: Use default settings for the Chunking and parsing configurations section, unless you would like to choose a chunking strategy as per your use case.
  3. Then you would choose from the available list of Embeddings model in Amazon Bedrock as per your requirement, and keep the default Vector dimensions of the chosen model.
  4. Next, choose from the available list of vector stores in the Vector database section, and create the knowledge base.
  5. Then you would test the knowledge base against your data by sending a prompt-query in the Amazon Bedrock knowledge base console. The query will validate against the data source you created as a knowledge base, and get a relevant response.
    注: See Test your knowledge base .
  6. As a logical next step, you would now navigate to the Automation Anywhere Control Room and navigate to AI > Model connections > Create Model connection to begin creating one.
  7. Use the Knowledge base ID from Amazon Bedrock to create the モデル接続 in Authentication details > Knowledge base ID.

    Adding this ID ensures that the Amazon Bedrock model references the specific content when retrieving a response to ensure response relevance and accuracy.

    See: Create Grounded モデル接続 with Amazon Bedrock RAG capability.

Refer to the following resources for a better understanding of how アマゾン ナレッジ ベース works and the steps to create a knowledge base: