Amazon.ae services around 10 million customers monthly in the Middle East and North Africa, spanning countries like the UAE, Saudi Arabia, and Egypt. Their AMET Payments team recently developed SAARAM, a multi-agent AI solution that dramatically decreases test case generation time from one week to mere hours. Traditional methods required extensive manual testing efforts, consuming valuable engineer resources and risking slow turnaround times. SAARAM leverages Amazon Bedrock with Claude Sonnet by Anthropic and the Strands Agents SDK to enhance quality and reduce errors by emulating human cognitive processes. The architecture efficiently orchestrates specialized agents, improving test coverage and standardizing testing across diverse regulatory environments. Key results include a 40% increase in edge case identification and improved payment success rates, which in turn enhanced customer satisfaction. With SAARAM’s implementation, the team is poised for future scalability across Amazon’s QA departments, exemplifying how strategic AI integration can transform traditional quality assurance methods.
Source link
Share
Read more