AWS Well-Architected Agent enters preview as the AI successor to Trusted Advisor

AWS has turned one of its oldest advisory products into an agent. The Well-Architected Agent, in public preview from October 1, scans a customer's AWS environment against the familiar Well-Architected pillars of cost, security, performance and reliability. Unlike Trusted Advisor's static rule checks, it correlates live application metrics and ranks findings against the customer's stated business goals.

The key change is moving from advice to action. Trusted Advisor flags an underused instance or an open security group, then leaves the work to an engineer. The new agent produces ready-to-implement fixes, which in practice means config changes or infrastructure-as-code patches that a human can review and apply. This matches the pattern in AWS's other October 1 posts: AWS Professional Services described a multi-agent migration system on Bedrock AgentCore that cuts IaC development from weeks to minutes, and a guide showed 'ambient' agents triggered by S3 uploads or alerts with an ask_human review step.

Competitively, Microsoft's Azure Advisor and Google's Recommender offer similar static guidance, so AWS is first among the hyperscalers to market an agentic replacement. It also gives AWS a reason for customers to stay inside its tooling rather than buy third-party FinOps or CSPM products, many of which are adding AI fixers of their own.

Caveats: the preview's pricing, regions and whether the agent can apply changes autonomously, rather than only proposing them, are not detailed in the sources. Given this week's rogue-agent headlines, enterprises will want tight IAM scoping before letting any agent edit production. Watch for GA pricing and integration with AWS's new agent-first developer onboarding.