AWS previews FinOps Agent to automate cloud cost analysis

AWS is pushing cloud cost management from monitoring into action. Its new FinOps Agent, now in preview, can answer cost questions, generate recurring reports, investigate anomalies, surface optimization recommendations and push findings into Slack or Jira, a sign that spend governance is becoming part of everyday workflow rather than a monthly review.

For engineering managers, the point is not just that the agent can find idle resources or flag rightsizing and Savings Plans opportunities. AWS says it can run on a schedule, trigger when an anomaly is detected, or respond when an engineer asks a cost question, which means cost checks can become a routine part of delivery instead of a separate finance chore. AWS documentation says the tool is free during preview, though customers still pay standard per-request charges for the AWS API calls underneath.

That matters at a moment when AWS is also trying to prove that AI can improve operations at scale. In a June 10 post, the company said a six-engineer team rebuilt the Amazon Bedrock inference engine in 76 days, even though the project had originally been scoped for 30 developers over 12 to 18 months. AWS said structured pilots with Amazon Stores teams produced a median 4.5x gain in normalized deployment velocity, and some groups cleared 10x. One example, Perfect Order Experience, moved from a two-week cycle to shipping in an afternoon.

The message for product and engineering teams at monday.com is straightforward: buyers are no longer impressed by AI that only generates content. They want AI that resolves a specific operational pain, whether that is a cost anomaly, a recurring report, or a ticket that routes itself to the right owner. That is the same logic behind Amazon Bedrock Flows, which AWS says lets teams visually connect prompts, agents, knowledge bases, guardrails, Lambda and other services to build AI workflows.

AWS first added AI-assisted cost optimization to Amazon Q Developer in June 2025, including rightsizing, Savings Plans, Reserved Instances and idle-resource recommendations. The FinOps community’s 2025 FinOps X announcements later highlighted that move as a natural-language interface for FinOps work, reinforcing how quickly cost control is being folded into the same AI layer as coding and deployment.

For monday.com, the operational lesson is already visible in its broader ecosystem. An Umbrella case study says monday.com used automated recommendations and ticket creation to streamline FinOps workflows, a reminder that the company’s own customers are already applying the same pressure: show how AI speeds delivery, but also show how it keeps budgets, reliability and ownership visible.