Amazon AMET Payments: How Strands AI Agents Reduced Test Case Generation from Weeks to Hours

The Amazon AMET Payments team, serving approximately 10 million customers across five countries in the Middle East and North Africa, has reduced test case generation time from one week to mere hours. This was made possible by implementing the multi-agent AI solution SAARAM, which also improved the quality of test coverage.

If routine quality checks are taking weeks for your QA engineers, this is already a prime case for automation. Message our manager — he will run a free analysis of your business and niche and show exactly how to get a real business result from an AI agent in your case, not a nice-looking picture. Message the manager

The Problem: A Week of Manual Effort per Test Case

The AMET Payments team releases an average of five new features monthly. Each feature requires comprehensive test case generation, which traditionally consumed one week of manual effort per project. Quality assurance engineers spent this time analyzing business requirement documents (BRDs), design documents, UI mocks, and historical test preparations—a process that required one full-time engineer annually merely for test case creation.

Initial attempts using single-agent AI systems often produced generic outputs like “verify payment works correctly” instead of the specific, actionable test cases the QA team required. For example, specific details were needed, such as “verify that when a UAE customer selects cash on delivery (COD) for an order above 1,000 AED with a saved credit card, the system displays the COD fee of 11 AED and processes the payment through the COD gateway with order state transitioning to ‘pending delivery.’”

The Solution: SAARAM Multi-Agent System

Amazon developed SAARAM (QA Lifecycle App), a multi-agent AI solution that helps reduce test case generation from one week to hours. SAARAM is built using Amazon Bedrock with Claude Sonnet by Anthropic and the Strands Agents SDK.

A key breakthrough came from a human-centric approach: instead of asking “How should AI think about testing?”, the team asked “How do experienced humans think about testing?”. This led to detailed research into the cognitive workflows of senior QA professionals, who do not process documents holistically but work through specialized mental phases.

SAARAM consists of specialized agents, each focusing on a specific aspect of the testing process, mimicking expert approaches:

• Customer Segment Creator Agent. Identifies product user segments, creates decision matrices, and develops end-to-end scenarios.
• User Journey Mapper Agent. Builds flow and sequence diagrams, generates end-to-end scenarios, and detailed test steps.
• Customer Segment x Journey Coverage Agent. Combines inputs from previous agents to create detailed segment-specific analyses.
• State Transition Agent. Analyzes various product state points in customer journey flows, creating state diagrams and generating related test scenarios.

The system went through several iterations to overcome context length limitations, reduce hallucinations, and ensure scalability.

Results and Outlook

The implementation of SAARAM led to the following key results:

• Reduced Test Case Generation Time. From one week to mere hours.
• Improved Test Coverage Quality. AI agents generate specific and actionable test cases, reducing errors and omissions.
• Capture of Institutional Knowledge. The system helps standardize testing approaches and preserve the expertise of experienced testers.

The solution is already being used by the AMET QA team and is positioned for expansion across other QA teams in the International Emerging Stores and Payments (IESP) Org.

Source: aws.amazon.com

Want to learn how AI agents can accelerate your testing processes or other routine tasks? Message the manager. We will analyze your case for free and suggest the optimal solution.