Modern DevOps teams face growing complexity, frequent incidents, and constant pressure to improve reliability while doing more with fewer people. To address this, we can delegate operational tasks to autonomous serverless agents acting as an AI DevOps engineer inside our AWS environments.

In this talk, I will show how to build serverless agents using AWS Lambda, Amazon Bedrock, and Step Functions that can analyze infrastructure, review deployments, generate execution logs with findings and metrics, and automate incident reports and post-mortem analysis. These agents can continuously produce insights, highlight risks, and suggest improvements aligned with an infrastructure roadmap.