Knowledge Bases for Amazon Bedrock enable foundation models (FMs) and agents to retrieve contextual information from private data sources, enhancing the relevance, accuracy, and customisation of Retrieval Augmented Generation (RAG) responses. Amazon Bedrock has continuously expanded the choices of embedding models, vector stores, and FMs available for Knowledge Bases.
New data source connectors
In this latest update, Amazon Bedrock introduces support for additional data source connectors in preview, including web domains, Confluence, Salesforce, and SharePoint. These new connectors allow developers to integrate a wider range of enterprise data sources into their RAG applications, further expanding the breadth and depth of information available to FMs and agents.
Integrating web domains
By adding web domains as a data source, RAG applications can now access public data, such as company social media feeds, to improve the relevance, timeliness, and comprehensiveness of responses to user inputs. The blog post provides a step-by-step guide on how to configure a web crawler data source within a Knowledge Base.
Connecting Confluence, Salesforce, and SharePoint
The new connectors also enable the integration of popular enterprise content management systems, such as Confluence, Salesforce, and SharePoint, as data sources for Knowledge Bases. The post demonstrates the process of setting up a Confluence data source, including authentication methods and metadata configuration options.
Programmatic access and customisation
In addition to the console-based setup, the blog post highlights the ability to create data sources programmatically using the AWS Command Line Interface (AWS CLI) or AWS SDKs. This allows for greater flexibility and automation in the deployment and management of Knowledge Bases.
To conclude
The expansion of data source connectors for Knowledge Bases for Amazon Bedrock empowers developers to build more comprehensive and contextually relevant RAG applications. By integrating a wider range of enterprise data sources, including web domains, Confluence, Salesforce, and SharePoint, organisations can leverage the power of generative AI to deliver more accurate and customised responses to user queries. This latest update further strengthens the capabilities of Amazon Bedrock, positioning it as a leading platform for the development of advanced, data-driven generative AI solutions.”