AI Agents in Your Terraform: Guardrails That Actually Work

Handing infrastructure to AI is no longer a thought experiment, it is a daily decision for DevOps, platform, and cloud engineering teams. Today, Derek - DevSecOps engineer and trainer at More than Certified - breaks down what it actually takes to let agents near production, including where the guardrails belong, what determinism really costs you, and how to decide when a human still has to review the change. We discuss: Why models will never be deterministic, and why that is the wrong problem to fix Where guardrails actually belong: pre-commit hooks, policy as code, and CI Why he now merges Terraform after the apply instead of before How blast radius decides when a pull request needs a human How to keep secrets out of state when an agent has access to your workspace Who is accountable when an autonomous agent breaks production What AI changes about how teams plan, review, and hand out work Derek trains engineers on DevOps and security, and has spent the past couple of years building agents across his own platform, from chat and grading to lab and course building. Brought to you by Brainboard, your AI driven platform to visually design and manage cloud infrastructure, collaboratively: https://www.brainboard.co?utm_source=...