AI agents are your new infra users
2026-05-08
AI agents are your new infra users
Hey there! Welcome to Platform Weekly. Your weekly slice of platform engineering pie. Every week, we dive into another corner of the platform engineering universe, from industry insights and community news to practical lessons and live demos.
Plus… 80% (so far) are seeing positive ROI from AI!? We’re at the beginning of our State of AI in Platform Engineering survey. Share where your ROI stands!
The panel topic at next week's IaCConf has been keeping me up at night. "Platform Engineering in the Age of AI Teammates. Building platforms when AI agents are the new consumers."
Let’s pause on that for a second. AI agents are the new consumers.
Almost every platform engineering decision we've made in the last decade has been for human developers. Just as we began the journey of massively expanding the platform persona to security, observability, and beyond, we need to expand 10x further as soon the majority of platform use will be from AI agents.
And yeah… it works. Until it doesn’t. We’ve got the speed. But we’ve also get the weird breakage you can’t actually explain. We’ve 10x’d the ability to move fast, and thus 10x the power to break things.
The Spacelift team is running IaCConf 2026 on May 14 with exactly this question at the center. 20+ IaC leaders across 13 sessions, including Corey Quinn ("AI Speaks Terraform Like a Tourist"), Matt Gowie ("From IaC to Agents"), Amin Astaneh ("10x Velocity, Operational Risk"), and yours truly on the AI Teammates panel.
The questions worth your morning:
When does AI-generated IaC help vs hurt? 10x throughput is real (the gain is parallel agents, not faster individuals). But so is 10x blast radius.
What replaces Terraform workflows at scale? When agents skip the plan/review step, what catches them?
How do you deploy agents without losing control? Policy-as-code stops being a nice-to-have.
What does platform engineering look like for non-human users? Spoiler: it isn't a portal;)
Here's the angle I'll be bringing. This is a platform race, not a model race. Platform engineering for AI agents isn't a UX problem. It's a contract problem. Human developers needed ergonomic interfaces, helpful errors, opinionated defaults. Agents need deterministic feedback loops, structured failures, and blast-radius controls baked into the platform itself. The IDP that won the human-dev era doesn't disappear. It evolves into an Agentic Developer Platform (ADP) for the coming autonomous-agent era. Same backbone, new contract.
And IaC is the canary. It's where the agent contact surface is most direct, where guardrails are most testable, and where the consequences are most immediate. Get the platform contract right, your throughput multiplies. Get it wrong, you're the disaster girl meme.
Except it's 15,000 resources, not 150
Register now. Virtual, free, and if you're sitting on this question, May 14 is where we dive into some answers!
Quick bites
Highlight of the week:
From the community:
Want to watch the recordings of PlatformCon workshops and talks you weren't able to attend? We're uploading more every day on the community YouTube channel.