How the Amazon AMET Payments team accelerates test case generation with Strands Agents

This article describes how Amazon's AMET Payments team built SAARAM, a multi-agent AI solution using Amazon Bedrock and Strands Agents SDK to accelerate test case generation for payment features across the Middle East and North Africa region.

• Reduced test case generation time from 1 week to hours using multi-agent architecture
• Studied human QA expert cognitive patterns instead of optimizing AI algorithms alone
• Implemented five specialized agents: Customer Segment Creator, User Journey Mapper, Coverage Analyzer, State Transition Agent, and Test Generator
• Used Pydantic structured outputs to reduce hallucinations and ensure type-safe responses
• Achieved 40% more edge cases identified and improved from 1.0 FTE to 0.2 FTE for validation
• Leveraged Strands Agents workflow orchestration for parallel and sequential task execution
• Integrated Amazon Bedrock with Claude Sonnet for sophisticated reasoning capabilities
• Plans future expansion using Amazon Bedrock Knowledge Bases and AgentCore for production deployment

SAARAM demonstrates how human-centric AI design combined with multi-agent orchestration can enhance QA processes while maintaining quality standards and reducing manual effort significantly.