AWS launches knowledge retrieval agent solution with Amazon Bedrock and CDK

AWS launches a knowledge retrieval agent solution with Amazon Bedrock and CDK. The new approach enables the automation of complex workflows with multiple data sources.

The solution includes two knowledge bases, each with an AgenticRetrieveStream, which enable multi-step search and grounded answers. It is implemented as a single CDK application in Python, with all resources provisioned in a single command line. The retrieval is agentic, which means it can make decisions and take actions based on data dynamically.

The solution is based on technologies such as Cohere Embed v4 and Amazon OpenSearch Service, which ensure fast and accurate search. Amazon Bedrock, together with the AWS CDK, offers new knowledge retrieval agent solutions, enabling automation and efficiency in text generation systems. Organizations such as Bayer Crop Science, Octave and Denodo have already implemented solutions based on these capabilities, integrating data from multiple sources and creating secure, scalable architectures.

Bayer developed the Agentic Decision Engine (ADE), an agent-based system to automate and explain machine learning workflows. The ADE integrates data from multiple sources, such as CDE, and delivers answers in up to 10 minutes. The ADE has 22 specialized agents and uses LangGraph for flow control. Octave Reliance built a multi-tenant AI database using Amazon Bedrock to meet data isolation requirements.

AWS redefines knowledge retrieval with agents that run in multiple steps, integrating information and automating complex workflows. Bayer Crop Science and Octave Reliance have already tested similar versions with Amazon Bedrock and CDK, demonstrating the model's scalability and efficiency. Availability in Brazil has not yet been confirmed, but the technology is already in use in other markets.

More images

Text written with AI from the sources listed and checked by the newsroom's automated rules. How we work

• #software