Amazon Bedrock Agents, a generative artificial intelligence (AI) application, has introduced two new managed capabilities in a preview version. The first capability is the ability for agents to retain memory throughout multiple interactions with users. This feature allows agents to remember previous conversations and user preferences, creating a personalized and efficient experience, especially for complex multi-step tasks like booking flights or processing insurance claims.
The second capability is support for code interpretation, enabling agents to dynamically generate and execute code fragments within a secure environment. This feature expands the usability of agents to handle complex tasks such as data analysis, text processing, and equation solving. With the ability to process various types of input files, agents can now also generate visualizations and graphs, making data interpretation more accessible.
To demonstrate these new capabilities, users can create agents using the Amazon Bedrock Console and utilize features like memory retention to create personalized experiences, or code interpretation to handle data analysis and visualization tasks. For example, users can ask the agent to book flights, calculate mathematical operations, or analyze data from attached documents.
The retention of memory and code interpretation features provide users with a seamless and adaptable experience, expanding the capabilities of Amazon Bedrock Agents. These functions are currently available in preview mode, and while there are no additional costs during the preview period, standard model usage charges apply. For more technical details and utilization examples, users can visit the Amazon Bedrock Agents section of the User Guide or explore the AWS community for further insights on using generative AI solutions.
Article Source
https://aws.amazon.com/blogs/aws/agents-for-amazon-bedrock-now-support-memory-retention-and-code-interpretation-preview/