AWS expands agentic automation with Nova Act QA testing and a DevOps Agent for incident remediation

AWS rolled out a coordinated set of agentic-automation tools aimed at the software-delivery lifecycle. Amazon Nova Act QA Studio expanded from its foundation to batch regression testing and CI/CD pipeline integration (Part 2), letting test suites organize and parallelize execution while a CLI brings agentic testing directly into automated pipelines. A companion release scales UX testing by auto-generating test scenarios from documentation and executing user flows via Nova Act's navigation.

The most operationally consequential piece is the AWS DevOps Agent paired with Kiro CLI, which performs automated incident remediation — turning investigation findings into deployed fixes without manual toil. It correlates telemetry across CloudWatch, deployment pipelines, and application logs, then writes and deploys the fix, explicitly targeting the late-night on-call burden. That's a step beyond 'observability with AI' toward AI that closes the loop autonomously.

AWS also addressed the flip side of the agent boom: AWS WAF Bot Control now helps distinguish legitimate AI-agent traffic from malicious bots. Because IP-based filtering fails in multi-tenant systems like Bedrock AgentCore where thousands of workloads share IP space, the new approach verifies agent identity instead — a recognition that AI agents are becoming first-class web clients.

The competitive context: this is AWS racing Microsoft (Copilot/GitHub) and Google to own the agentic DevOps and QA layer, extending its Bedrock AgentCore platform. The skeptical caveat is that auto-deploying AI-written fixes into production is exactly where trust breaks down — a wrong 'remediation' at 2am could cause the outage it was meant to fix. Watch adoption among risk-averse enterprise ops teams and what guardrails AWS mandates.