Question

A developer builds a document-analysis feature where an API Gateway REST endpoint triggers a Lambda function that calls Bedrock's InvokeModel with a large Claude model to summarize lengthy contracts. Under load, some summaries take 45-90 seconds to generate, and clients receive HTTP 504 errors. The team wants a reliable pattern that returns quickly to the caller while still delivering the completed summary. Which change best resolves this?

Accepted Answer

Convert the workflow to an asynchronous pattern: API Gateway accepts the request and returns a job ID, a Step Functions Standard workflow invokes Bedrock, and the client polls a status endpoint or receives a callback.

Answer

Increase the Lambda function timeout to 15 minutes and raise the API Gateway integration timeout beyond its 29-second maximum.

Answer

Switch to the Bedrock InvokeModelWithResponseStream API from within the same synchronous Lambda so tokens stream back through API Gateway before the timeout.

Answer

Enable Provisioned Throughput for the Claude model so that every summarization completes in under 29 seconds regardless of contract length.