Confidently Wrong: An AI-Era Infrastructure Risk Workshop for DevOps & SRE Teams
AI-assisted engineering is changing how code, infrastructure, and operational decisions get made, and the failure mode is rarely a change that looks wrong. It is a change that looks finished: a Terraform plan that quietly widens an IAM policy, a Kubernetes manifest that drops a resource limit, a refactor that replaces a database instead of updating it. Each one arrives with the same fluent confidence as the correct version.
This workshop is built around those changes. Teams work through a set of intentionally flawed, AI-generated Terraform/OpenTofu and Kubernetes examples, find the problems themselves, and then map where AI tools, CI/CD, permissions, secrets, and automation increase leverage and blast radius in their own environment. It shares its material with the book Confidently Wrong and with an open-source plan checker, so what the team practises in the room is what their pipeline can enforce afterwards. Practical clarity, not an AI hype session.
What we cover
Private, customized to your stack and team — not a canned training deck.
Topics we work through
- Hands-on: reviewing flawed AI-generated Terraform and Kubernetes changes
- The plan signals that deserve a human: IAM expansion, public exposure, replacement of stateful resources
- AI agents as operational actors
- CI/CD as a control plane
- Permission boundaries & trust
- Human review that actually matters
- Guardrails, policy, and blast radius
- Practical next steps for the team
Your team leaves with
Practised review habits for AI-generated infrastructure changes, a clearer map of where AI-assisted work can affect production, and a prioritized list of guardrails, including checks your CI can run on every plan.
Who it’s for
DevOps, SRE, platform, and security-minded engineering teams adopting AI coding tools, agents, or CI/CD automation who want shared language on operational risk before it shows up in an incident or audit.
From intake to outcomes
How the session works
- Pre-session intake on your stack, team size, and the decisions in play
- Customized agenda built from the topic list — not a generic slide deck
- Interactive working time with your engineers on real examples and tradeoffs
- Clear outcomes and next steps your team can act on after the session
Pick a length that fits your team’s day
Most private sessions run as a 90-minute briefing or a half-day deep dive — multi-session series are available too.
90-Minute Briefing
Tight, high-density introduction. Great for engineering all-hands or leadership briefings.
Half-Day Workshop
Teaching, working examples, and team discussion. The most common private format.
Multi-Session Series
Spread across weeks for deeper coverage and time to apply between sessions.
Leadership + Engineering
Pair a leadership briefing with a follow-up team session for shared language.
This session is a fit if…
- Your team is adopting AI coding tools, agents, or heavier CI/CD automation
- Permissions and blast radius from the developer toolchain are hard to explain to leadership
- You want shared language on operational risk before an incident or audit
- Security-minded engineers and platform leads need the same mental model of what can reach prod
Plan Confidently Wrong: Using AI Safely in DevOps & SRE.
Tell me a bit about your stack, team size, and the format you want. I’ll come back with a proposal and timing — most sessions are available as a 90-minute briefing or a half-day deep dive.
Plan This Workshop → ← All workshops