Infrastructure-as-Code (IaC)

AI Agents Are Non-Deterministic. So Are You. Deal with It.

Let me start with a question. Do you like games? Video games, board games, whatever it is. And would you play them all day if you actually could? I know I would.

Here’s my problem. I can’t. I like games. But I also like money. Money for rent, for food, for more games. And to get that money, I have to get some shit done for the company that pays my bills. That’s the deal.

So my real dream was never “more games.” It’s getting the work done without me having to do it, so I can get back to the controller. There’s all kinds of work I’d happily hand off, but I want to zoom in on one slice of it: the ops work. Provisioning the infrastructure, wiring the databases together, deploying apps, keeping the whole thing running. That’s what I’m trying to get AI agents to do for me. Today.

And that dream is not crazy. It’s not just hype. An agent really can do that work. It provisions the clusters, wires up the databases, fixes the broken pipeline, and you get to lean back and pick up the controller.

But there’s a catch. And right about now, you’re probably yelling at the screen. In your head, you’re yelling at me. AI is not deterministic! Well, first of all, I can’t hear you, so please stop yelling at your monitor. And second, you’re right. That agent working for you is non-deterministic. Give it the same task twice and you can get two completely different answers. And it’ll do risky things with total confidence, whether it’s right or whether it’s dead wrong.

But sound familiar? Because that’s you too. Give the same task to two engineers, or the same engineer on two different days, and you get two different answers, each delivered with the exact same confidence.

The only real difference is that you are so annoyingly slow. You make the occasional mistake. An agent makes a thousand an hour. And it never gets tired.

One Control Plane for Every GPU Cluster (Modeplane)

We’ve been working on something new. A project called Modelplane. It’s early, it’s rough… but I think it’s ready to fly.

But before I show you what it does, let me back up and explain the problem it solves. Because that’s really where this whole thing starts.

Serving a single model on a single cluster is more or less a solved problem. Pick a serving engine, hand it a GPU, point some traffic at it, and you’re done. The hard version is serving models at scale. GPUs are scarce and expensive, and they’re scattered all over the place, across regions, across clouds, and across your own on-prem hardware, wherever you could actually get your hands on them. And the models people really care about, the big ones, won’t even fit on a single machine. So you don’t end up with a cluster. You end up with a whole fleet of GPU clusters.

Infrastructure with AI Agents for Dummies

AI agents are amplifiers. If you’re good at your job, agents make you better. You do more great things, faster. But if you’re bad at your job, agents amplify that too. Where you used to cause a slow trickle of shit, now you have the means to unleash a full-blown shitstorm, at scale, in minutes.

Now, AI is all the rage these days, and for good reason. So of course people are using agents to manage real resources: infrastructure, databases, applications, all of it. The question is what happens when they do. That’s what we’re looking at today: an agent managing actual cloud resources, what goes wrong, why it goes wrong, and what it takes to make it work properly.

The End of Infrastructure-as-Code: AI Changes Everything

Here’s something that might shock you: every Infrastructure-as-Code tool you’re using today will be dead in upcoming years. Terraform, Pulumi; all of them. And the killer? AI agents that don’t need the abstractions we’ve built our entire industry around.

I know this sounds crazy. You’re probably thinking, “But we just invested years learning these tools!” Here’s the thing though - AI agents don’t operate like humans. They don’t get overwhelmed by complexity in the same way we do. They don’t need user-friendly abstractions. And that changes everything.

By the end of this video, you’ll see exactly why the tools we consider essential today are actually obstacles for AI agents, and what’s coming to replace them. But first, let me show you how we got here…

AWS Controllers for Kubernetes (ACK): Is It Worth It?

It’s not a secret that I believe in Kubernetes and that I think that it is much more than a “thingy where we run containers”. It is an extensible API with controllers that can manage any type of resources. It is a control plane for everything.

That’s why, a while ago, I was very excited when AWS Controllers for Kubernetes (ACK) project was born. It allows us to extend Kubernetes with Custom Resource Definitions and controllers that allow us to manage AWS resources through Kubernetes.

Terraform vs. Crossplane vs. Ansible - Rivals or Allies?

I am often asked to compare Crossplane with Terraform, or Pulumi, or Ansible, or any other tool that primarily manages resources, be it those in hyperscalers like AWS, Google Cloud, and Azure, or in Kubernetes, or anywhere else. Well… Today I’m here to tell you that none of those tools are going away any time soon. We need all of those. We need configuration management tools like Ansible, we need Infrastructure-as-Code (IaC) tools like Terraform and Pulumi, and we need control planes, be it opinionated ones like AWS, Google Cloud, and Azure, or those that allow us to build our own control planes like Crossplane.