Skip to content

README.md

Latest commit

 

History

History
85 lines (62 loc) · 3.19 KB

README.md

File metadata and controls

85 lines (62 loc) · 3.19 KB
Raw

Grounded AI Answers with Amazon Bedrock AgentCore Web Search and Amazon Bedrock (CDK)

This pattern deploys an Amazon API Gateway REST API backed by an AWS Lambda function that searches the live web via Amazon Bedrock AgentCore Web Search, then uses Amazon Bedrock to synthesize accurate, cited answers grounded in current facts.

Learn more about this pattern at Serverless Land Patterns: https://serverlessland.com/patterns/apigw-lambda-websearch-bedrock-cdk

Important: this application uses various AWS services and there are costs associated with these services after the Free Tier usage - please see the AWS Pricing page for details.

Requirements

  • Create an AWS account if you do not already have one and log in.
  • AWS CLI installed and configured
  • Node.js 20+ installed
  • AWS CDK v2 installed (npm install -g aws-cdk)
  • CDK bootstrapped in your account/region (cdk bootstrap)
  • Amazon Bedrock model access enabled for Claude Sonnet 4 in us-east-1

Deployment Instructions

  1. Navigate to the pattern directory:

    cd apigw-lambda-websearch-bedrock-cdk

  2. Install dependencies:

    npm install

  3. Deploy:

    npx cdk deploy

How it works

This pattern demonstrates Retrieval-Augmented Generation (RAG) using live web data instead of a static vector database:

  1. Amazon API Gateway receives user questions via POST /ask
  2. AWS Lambda orchestrates the two-step process:
    • Searches the live web via Amazon Bedrock AgentCore Gateway's Web Search connector (MCP protocol + SigV4)
    • Passes search results as context to Amazon Bedrock (Claude Sonnet 4) for inference
  3. Amazon Bedrock synthesizes a grounded answer with numbered citation references
  4. Response includes the answer text and source URLs for verification

This eliminates the need for maintaining a vector database, embedding pipeline, or data ingestion — the web IS the knowledge base, always current.

Testing

curl -X POST https://YOUR_API_ID.execute-api.us-east-1.amazonaws.com/prod/ask \
  -H "Content-Type: application/json" \
  -d '{"question": "What was announced at AWS Summit NYC 2026?"}'

Or via AWS Lambda directly:

aws lambda invoke --function-name FUNCTION_NAME \
  --payload '{"body": "{\"question\": \"What is Amazon Aurora DSQL?\"}"}' \
  --cli-binary-format raw-in-base64-out output.json && cat output.json

Example Response

{
  "answer": "Amazon Aurora DSQL is a serverless, distributed SQL database [1] that provides active-active multi-region support with PostgreSQL compatibility [2]...",
  "sources": [
    {"title": "Amazon Aurora DSQL documentation", "url": "https://docs.aws.amazon.com/aurora-dsql/..."},
    {"title": "Aurora DSQL launch blog", "url": "https://aws.amazon.com/blogs/..."}
  ]
}

Cleanup

npx cdk destroy

Copyright 2026 Amazon.com, Inc. or its affiliates. All Rights Reserved.

SPDX-License-Identifier: MIT-0