Accelerating software delivery with agentic QA automation using Amazon Nova Act
New agentic service uses visual AI to automate testing without brittle code selectors, cutting maintenance overhead.
AWS has launched Amazon Nova Act, a new service designed to revolutionize quality assurance automation through agentic AI. The service employs a custom computer vision model that interacts with web applications the same way humans do—by seeing the screen and understanding context—rather than relying on fragile code selectors like XPaths or CSS identifiers. This visual approach allows automated tests to automatically adapt when developers refactor UI code or designers change layouts, addressing the primary maintenance burden that plagues traditional frameworks like Selenium. Teams can define test steps using natural language instructions that Nova Act translates into browser actions, creating a direct link between product requirements written in business language and executable test cases.
To operationalize this capability, AWS provides QA Studio, a serverless reference solution built on Nova Act that offers a web interface, API, and CLI for managing test suites. QA Studio enables test creation through live browser previews powered by Amazon Bedrock AgentCore Browser and can generate tests from user journey descriptions. The system provides full visibility into test execution with trajectory logs that show the AI's visual reasoning at each step, transforming debugging from parsing technical stack traces into understanding natural language descriptions of test behavior. This end-to-end approach allows teams to schedule tests, run them on-demand, or trigger them within CI/CD pipelines while focusing on feature delivery rather than test maintenance.
- Amazon Nova Act uses visual AI and natural language to automate QA, eliminating dependency on brittle code selectors that break with UI changes
- The service integrates with QA Studio, a serverless reference solution that provides test management, scheduling, and CI/CD pipeline integration
- Provides trajectory logs showing AI's visual reasoning at each step, making debugging transparent through natural language descriptions instead of technical stack traces
Why It Matters
Dramatically reduces QA maintenance overhead by up to 70%, accelerates release cycles, and allows non-technical team members to create and manage automated tests.