Rescale, a digital engineering platform provider, has announced the successful deployment of an automated AI triage agent designed to manage its internal software testing cycles. The system, built using the AI-Driven Development Lifecycle (AI-DLC) framework on AWS, identifies and routes failures in nightly end-to-end test suites, significantly reducing the manual operational burden on engineering teams.
Automating complex engineering workflows
Prior to the implementation, Rescale’s engineering teams faced repetitive, manual tasks to classify test failures before deployment. The new agent, integrated into the company’s existing Argo Workflows CI/CD pipeline on Amazon Elastic Kubernetes Service (Amazon EKS), automates this process by querying logs and historical data. When a failure occurs, the agent utilizes Claude Opus 4.8 via Amazon Bedrock to categorize the issue—whether it is a code bug, infrastructure failure, environment error, or a flaky test—and generates a structured remediation report.
Building with the AI-DLC framework
The development of the triage agent was conducted over a five-day workshop utilizing the AI-DLC methodology. This approach emphasizes collaborative “mob sessions” where humans define requirements, validate architectural decisions, and oversee AI-generated code. By adopting this structure, Rescale moved from initial problem definition to a staging deployment in five days, with a production release achieved within several weeks.
Key technical components of the solution include the use of the Strands SDK for agent orchestration, Amazon CloudWatch for log retrieval through the Model Context Protocol (MCP), and a dedicated knowledge base for tracking historical failure patterns.
Operational impact and adoption
Since its deployment, the triage agent has demonstrated a steady improvement in accuracy, with the classification agreement between the AI and human triage rising to 68%. The company reports a 20% reduction in time blocking production deployments related to test failure analysis, allowing engineers to dedicate more time to architectural tasks and product development.
Following the success of the triage project, Rescale is now extending the use of AI-DLC across other engineering initiatives, including its simulation-domain product development. The company is also evaluating the creation of an internal AI-DLC Center of Excellence to standardize these practices across its broader organization, marking a shift toward an AI-first approach to complex software engineering.
Source: aws.amazon.com
Original article: https://aws.amazon.com/blogs/machine-learning/how-rescale-built-an-ai-triage-agent-with-strands-agents-and-ai-dlc/
Leave a Reply