article --summary
The summary
This post shows how a financial services team can wire three AI agents into a single GitHub delivery pipeline while keeping human approval and audit evidence in place. Kiro writes the code, AWS DevOps Agent reviews release readiness before and during the pull request, and AWS Security Agent runs a deeper security pass on the PR. A person still approves the merge, and the same DevOps agent investigates production failures afterward.
The authors frame the problem around regulatory pressure to manage vulnerabilities quickly and continuously, and under DORA the institution remains accountable for ICT risk even when using cloud or third-party tooling. The reader learns the role each agent plays across the SDLC and why workflow-level orchestration helps keep context and evidence in one place.
The practical section covers prerequisites, including a GitHub repo, an AWS account with an OIDC provider, Kiro, and Slack, plus installing the GitHub app and the DevOps agent Kiro Power. It then steps through deploying a Python application to AWS Lambda with GitHub Actions. The authors stress that the reference implementation is not compliance on its own, so each institution must map it to its own policies and controls.
Editorial note: This is a summary of an article first published by AWS DevOps Blog. The full piece is on their site.