At AWS re:Invent, Amazon announced a number of new capabilities for generative AI and machine learning.

New evaluations for Amazon Bedrock

Two new evaluation capabilities have been added to Amazon Bedrock, which is the company’s platform for building generative AI applications using foundation models. 

Amazon Bedrock Knowledge Bases now offer support for RAG evaluation (still in preview), which allows users to compare different configurations. Amazon Bedrock Model Evaluation now offers an LLM-as-a-judge feature (also in preview), which allows AI to perform tests and evaluate other models more quickly and cheaper than a human. 

The evaluations provide a score of 0 to 1 and provide a natural language explanation for each score, as well as the rubric used to come to each score. 

“These new capabilities make it easier to go into production by providing fast, automated evaluation of AI-powered applications, shortening feedback loops and speeding up improvements. These evaluations assess multiple quality dimensions including correctness, helpfulness, and responsible AI criteria such as answer refusal and harmfulness,” Danilo Poccia, chief evangelist at AWS, wrote in a blog post. 

RELATED: Observability enhancements announced at AWS re:Invent

Amazon Bedrock gets new APIs for RAG applications

Another update to Amazon Bedrock is the addition of custom connectors and ingestion of streaming data in Amazon Bedrock Knowledge Bases.

“With support for custom connector and ingestion of streaming data, customers can now use direct APIs to efficiently add, check the status of, and delete data, without the need to list and sync the entire dataset,” Veliswa Boya, senior developer advocate at AWS, wrote in a blog post. 

PartyRock’s new capabilities

PartyRock, a playground for Amazon Bedrock, was given several updates as well. The company is eliminating the free trial period and instead offering all users a free daily usage allotment, starting next year. 

Another new feature is the ability to search the app catalog to find apps that fit a particular use case.

And finally, users can now upload and process multiple documents at once, which makes it easier to build apps that handle batch processing, document comparison, or content aggregation. 

Amazon Q Business can now leverage extensions and integrations

Amazon Q Business is a generative AI assistant for enterprise, and new browser extensions and integrations will enable it to be more helpful to users. 

It can now be added as an extension in the browser, which allows it to be used when visiting webpages for tasks like summarizing the content on a page. At launch, Chromium-based web browsers, including Google Chrome and Microsoft Edge, are supported, as is Mozilla Firefox. 

Amazon Q can also now be integrated into different platforms, such as Slack and Microsoft Teams. 

Finally, it can also now extract information from visual components, such as diagrams, infographics, charts, and other image-based content.