QuestionQ93

VerifiedOperational Efficiency and Optimization for GenAI Applications

An ecommerce company uses Amazon Bedrock to develop a generative AI (GenAI) application. The application uses AWS Step Functions to orchestrate a multi-agent workflow that produces detailed product descriptions. The workflow contains three sequential states: a description generator, a technical-specifications validator, and a brand-voice consistency checker. Each state generates intermediate reasoning traces and outputs that are passed to the following state. The application uses an Amazon S3 bucket for process storage and output storage.

During testing, the company finds that outputs passed between Step Functions states frequently exceed the 256 KB quota, causing workflow failures.

A GenAI Developer must revise the application architecture to efficiently handle the Step Functions 256 KB quota while maintaining workflow observability. The revised architecture must retain the existing multi-agent reasoning and acting (ReAct) pattern.

Which solution meets these requirements with the LEAST operational overhead?

Explanation

The optimized Amazon Bedrock integration for AWS Step Functions can read large model input from an Amazon S3 URI and write model output to an S3 URI. When an output URI is specified, the state receives an S3 reference instead of the full response body, keeping the data passed between sequential states below the 256 KiB limit. ResultSelector and ResultPath can preserve and route these references while retaining the sequential multi-agent workflow and its observable artifacts in S3.

Learn more

Community Discussion

No comments yet. Be the first to start the discussion!