AWS has published a reference implementation for synthetic monitoring with Amazon Nova Act and Bedrock AgentCore. Instead of checking only servers or individual APIs, the system repeatedly performs complete browser journeys such as signing in, searching for a product, adding it to a cart, and confirming checkout.
That end-to-end view can catch failures that infrastructure dashboards miss, including a broken frontend button, a changed third-party login page, or an unresponsive form. Traditional Selenium and Playwright tests often depend on exact Document Object Model selectors, which can fail after harmless layout changes. AWS’s approach uses an agent to interpret the page and carry out the intended task, while AgentCore supplies managed execution around the workflow.
The post includes architecture patterns and a sample repository rather than a turnkey guarantee of reliable monitoring. Agent-based tests can adapt to interface changes, but their probabilistic decisions also require clear success criteria, logs, and limits so a mistaken action is diagnosable. The pattern is relevant beyond ecommerce: booking, subscription, account-access, and appointment systems all have critical journeys whose health cannot be established from backend availability alone.