Amazon AMET Payments Accelerated Test Case Generation 20x: How Strands AI Agents Transformed QA Processes

The Amazon AMET Payments team, managing payment systems for 10 million customers across five countries, launches approximately five new features monthly. Each feature requires extensive testing, and traditionally, test case generation took a week of manual effort. After implementing a multi-agent AI system with Strands Agents, this time was cut down to just a few hours, a 20x improvement.

If your company spends a week on manual test case generation, this is already a prime candidate 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

How It Was Before AI Agents

Amazon AMET Payments QA engineers spent a full week on each project to manually analyze business requirements, design documents, UI mockups, and past test preparations. This process was so time-consuming that test case generation alone required one full-time engineer annually.

Key challenges included:

• High time consumption. A week for test case generation for each new feature.
• Repetitive tasks. Engineers spent time on routine preparation and documentation instead of strategic testing initiatives.
• Scaling difficulties. With five new features per month, this approach created a bottleneck in the product development cycle.

Initial attempts using traditional AI approaches, by simply feeding entire documents to a single agent, yielded overly generic results, such as “verify payment works correctly,” instead of specific and actionable test cases.

What Was Implemented: SAARAM and Multi-Agent Approach

The team developed SAARAM (QA Lifecycle App), a multi-agent AI solution utilizing Amazon Bedrock and Anthropic’s Claude Sonnet combined with the Strands Agents SDK. The key was 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 the creation of specialized agents that mimic the cognitive processes of experienced QA engineers:

• Document analysis. Agents extract acceptance criteria, identify customer journeys, analyze UX requirements, and user data.
• Test development. A systematic process including journey analysis, scenario identification, data flow mapping, and test case development.

Iterations involved creating specialized agents for customer segmentation, user journey mapping, segment coverage analysis, and state transition management. The Strands Agents SDK allowed for efficient orchestration of these complex, interdependent tasks.

Results

The implementation of SAARAM significantly accelerated the test case generation process, improved its quality, and reduced costs. This allowed QA engineers to focus on more strategic tasks rather than routine work. The solution is scalable and planned for expansion to other Amazon QA teams.

Source: aws.amazon.com

What You Can Do

Amazon AMET Payments’ experience demonstrates that multi-agent AI systems can transform processes requiring deep analysis and systematization:

• Automate routine analysis. AI agents can analyze large volumes of documentation, extracting key information and structuring it for further use.
• Generate specific tasks. Instead of general statements, agents can create concrete, actionable tasks by mimicking the thinking of experienced specialists.
• Scale expertise. The knowledge of top employees can be “encoded” into agents, allowing their experience to be scaled across the entire team and reducing reliance on individual experts.

If you want to learn how AI agents can optimize processes in your company, message the manager. We will conduct a free analysis and show the potential for your business.