Streamline grant proposal reviews using Amazon Bedrock (AWS Health Equity Initiative)

Use case typeLegal document automationUpdated Jun 13, 2026

The AWS Social Responsibility & Impact team built a prototype to streamline grant proposal review and evaluation for the AWS Health Equity Initiative. The solution uses Amazon Bedrock with Claude 3 Sonnet, structured personas and rubrics, and a Streamlit interface to produce JSON assessments for grant submissions.

Organization
AWS Social Responsibility & Impact
Industry
Public Sector
Location
United States
Published
January 2025

Reported outcomes

2 days

timeTime & speed

Strategic outcomes

Speed & agilityStreamlined grant proposal reviewsBetter decisions & insightStandardized proposal evaluationBetter decisions & insightCaptured assessment data for analysisCustomer experience & trustMore consistent evaluator assessments

Primary read

Use case focus

Showing 3 of 3

  • 1Document Review and Analysis
  • 2Workflow Automation
  • 3Decision Support
Manual review of grant proposals took 14+ days per cycle and became difficult to scale as application volume increased to 139 submissions in one cycle.
  • The team created a Streamlit prototype that lets reviewers select personas and rubrics, dynamically constructs prompts, and sends them to Amazon Bedrock using Claude 3 Sonnet.
  • Review input and generated assessments are stored in DynamoDB, with AWS Lambda and Amazon S3 included in the surrounding architecture and serverless scalability considerations.
  • The approach standardizes proposal evaluation by asking the model to return structured JSON assessments across rubric dimensions.
Technologies
Amazon BedrockAnthropic Claude 3.5 SonnetAmazon DynamoDBAWS LambdaAmazon S3StreamlitAWS Bedrock Runtime
  • Reduced review time from 14+ days to about 2 days.
  • Enabled structured capture of proposal assessment data for further analysis.
  • Supported more consistent evaluation across multiple personas and rubrics.
Architecture

A Streamlit web prototype stores personas, rubrics, and submissions in DynamoDB, constructs dynamic prompts from selected persona and rubric data, and calls Amazon Bedrock runtime with Claude 3 Sonnet to generate structured JSON assessments. The article also discusses production considerations using API Gateway, S3, KMS, IAM, CloudWatch, CodePipeline, and Bedrock Guardrails.

Sources & evidence1
Groundedness: 4/5Type: Blog PostPublished: Jan 30, 2025Publisher: AWSEvidence: VendorConfidence: Medium
Primary source

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases
Customer evidence
More from AWS Social Responsibility & Impact
Provider evidence
AWS provider pageAWS ranked cases
Market context
Public Sector AI use casesPublic Sector industry insightsUnited States country insights
Browse the catalog
All AI use casesAI use-case typesLegal document automation use cases

Was this useful?