How Rescale built an AI triage agent with Strands Agents and AI-DLC

What happened

Rescale used Strands Agents and the AI-DLC framework on AWS to build a test-failure triage agent, reaching staging in five days and production within weeks. The agent classifies failures, retrieves historical context and logs, and recommends fixes for engineer review. Rescale reports classification agreement with human triage improving from 44% to 68% and a 20% decrease in time blocking production deployment while test failures are analyzed.

Read next

• More changes in today’s briefingMicrosoft Research and Collaborator Resolve the KLS Conjecture With Human-Directed AI

The story so far

Rescale built and deployed an AI agent to triage test failures using Strands Agents and AWS’s AI-Driven Development Lifecycle (AI-DLC) framework, reaching staging in five days and production within weeks. Rescale reports that the agent’s classification agreement with human triage improved from 44% to 68%, alongside a 20% decrease in the time test-failure analysis blocked production deployment. The agent classifies failures, retrieves historical context and logs, and recommends fixes for engineers to review. Its role is to support engineers investigating failed tests, with proposed fixes subject to human review.