Compliance audits demand meticulous documentation, often requiring teams to manually gather hundreds of screenshots across disparate systems like GitHub, AWS, and internal applications. This process is not only time-intensive but also prone to errors and inconsistencies, complicating reproducibility across audit cycles. Amazon Bedrock and browser automation now offer a solution to automate these workflows, reducing manual effort while ensuring accuracy and scalability.
A new system leverages Amazon Nova 2 Lite, an AI model integrated with a browser extension for Chrome and Firefox, to automate evidence collection. The extension executes predefined compliance workflows, captures timestamped screenshots, and stores them in Amazon S3, while also analyzing documents to generate workflows via natural language processing (NLP). Post-execution, Amazon Simple Email Service (SES) delivers compliance reports directly to specified email addresses.
The architecture combines browser automation with AI for three key advantages: universal compatibility with any web application, visual evidence capture for auditors, and adaptability to UI changes through intelligent automation. The system’s modular design includes a UI layer for user interaction, an AI agent layer powered by Amazon Nova 2 Lite for chat, workflow design, and report generation, and a workflow engine that processes JSON-defined steps with error recovery. Amazon Cognito ensures secure authentication, while AWS Lambda functions manage deployment prompts and bucket cleanup.
Evidence is organized in Amazon S3 with a structured hierarchy, storing screenshots, workflows, chat logs, and reports by date. The system’s audit logs and encryption at rest further enhance security and compliance. Organizations can deploy this solution to reduce manual labor, minimize errors, and streamline audit processes across multiple systems.
Read more: aws.amazon.com