We spent a decade managing cloud infrastructure for engineering teams. Not advising from the outside — actually in the weeds, building production systems for teams that couldn't afford a dedicated DevOps hire. We saw the same dysfunction, over and over, across dozens of companies.
The CTO handling infrastructure on the side. The stack that grew without a process. The change that went out on a Friday because someone assumed it was fine. We rebuilt the same setups from scratch for every new customer. At some point, we got lazy. What if we just built the platform once, and gave every team the infrastructure they'd get from hiring an agency — on their own account, at a fraction of the cost?
So we started talking to founders. And they weren't excited about creation at all. They already had infrastructure. What was killing them was everything after — the changes, the approvals that lived in someone's head, the incident that happened because a fix got applied without anyone checking blast radius. Creation was a one-time pain. Operations never ended.
“The problem was never that teams couldn't create infrastructure. It's that nobody knows if the next change is safe to apply.”
That insight cracked the whole problem open. And then AI arrived and made it a crisis.
Today any engineer — or any AI agent — can generate production-ready infrastructure code in minutes. The barrier to creating a change has collapsed completely. The barrier to making that change safe has not moved at all. If anything it's gotten worse, because now the volume of changes is exploding while the governance layer is still just a judgment call and a prayer — whether you are a two-person startup or a hundred-person engineering org.
Look at how most teams actually operate, regardless of size. Someone proposes a change. It gets applied whenever someone gets around to it, with whatever checks exist in someone's head. No governed lifecycle. No structured approval. No evidence trail. Lean startups feel this because there is no process at all. Larger teams feel it differently — buried under tickets, tribal knowledge, and approval chains that slow everything down without actually making anything safer. The dysfunction looks different at different scales. The root cause is identical.
The industry responded by building better tools. Better IaC. Better pipelines. Better policy engines. All of it misses the point. The problem was never the tool. It was the absence of a governed lifecycle around every change. A layer that sits between intent and production. That runs checks automatically. That routes approvals to the right person. That records every decision. That works regardless of whether the change was written by a human, pushed by a pipeline, or generated by an AI agent.
That layer doesn't exist. So we are building it.
We call it the Change Control Plane. Every infrastructure change — regardless of source, regardless of who made it, regardless of how large the team is — moves through the same governed path: propose, check, approve, apply, record. The atomic unit is a ChangeSet. Every ChangeSet carries its full context: what changed, why, who approved it, what the checks found, what got applied. Nothing reaches production without moving through that lifecycle.
We are starting with the teams that feel this most acutely today — lean engineering teams on AWS where the CTO is still the de facto infrastructure function and there is no platform team to fall back on. That is the wedge. But the governed change layer is not a workaround for teams without DevOps. It is the missing layer for everyone. The startup shipping its first production workload. The scale-up onboarding AI agents into its infrastructure workflow. The enterprise trying to maintain control as its cloud footprint sprawls across a hundred services and three providers.
The destination is a world where every infrastructure change — on any cloud, from any source, at any scale — moves through a governed lifecycle by default. Not as a compliance requirement. Not as an enterprise checkbox. As the only rational way to run infrastructure when the speed of change has outpaced any human's ability to review it manually.
Cloud infrastructure has been tool-obsessed for twenty years. Which IaC framework. Which pipeline. Which scanner. AI is making those questions irrelevant fast. The question that survives — the one that becomes more critical every time an agent touches your infrastructure — is brutally simple:
What change do we want to make, and is it safe to apply?
Everything we are building is an answer to that question.
CEO & Co-founder
CTO & Co-founder