Hey all,

I have a confession to make:

Spacelift didn’t start with some grand vision I had. It started with an insanely expensive mistake I made as a consultant.

Here’s the story:

After spending years as an engineer at Google and Facebook, I ended up consulting for scale-ups across Europe, helping them survive those moments when startup infrastructure becomes a liability.

One of my clients, Deliveroo, was scaling fast, and its underlying infrastructure strained under the weight. I came in to bring some method to the madness: tooling, process, change management, observability.

Infrastructure as code was a big part of that. I introduced Terraform, and then I introduced Terraform Enterprise, which cost hundreds of thousands per year.

It was supposed to make Terraform workable at scale. Instead, it caused a massive outage, deleting their infrastructure left, right and center.

Blaming Myself

To fix the problem, I first went to HashiCorp. I expected urgency, or at least an apology. But their response was basically, “Yeah, we know. We’ll fix it eventually.” Meanwhile, I had just helped a client pay a fortune for a tool that:

1. Didn’t solve the actual problem.

2. Tried to eat their infrastructure.

My manager didn’t blame me, but I blamed myself. So I did what engineers do when they feel guilty:

I opened a laptop.

Terraform itself is a fantastic single-player tool. But when multiple users touch the same infrastructure, it needs guardrails, coordination and an auditable workflow. So I pulled in someone I’d met at a few meetups (he’s still at Spacelift today!), and we built an on-prem replacement for those pieces of Terraform Enterprise that Deliveroo actually needed.

They started using it. And they loved it. Naturally, my next thought was:

Maybe HashiCorp wants to see this.

‘This Is Bollocks’

So I went back to them. Not with a pitch deck, but with a job application. I’d seen they were hiring a Terraform Enterprise product manager. At my interview, I said: 

“I love your company. I love your product. But I just had to rip Terraform Enterprise out of a production environment and replace it with something I built. I’m happy to walk you through it. Hire me, and I’ll help you fix it.”

They listened. They looked at what we built. And they pretty much told me, “This is bollocks. This isn’t the direction we’re going.”

Still, I didn’t walk out of that interview thinking, “I should start a company.” I went back to work.

Over time, though, people I worked with at Deliveroo joined other startups and scale-ups… and they kept coming back to me with the same request:

“Can we get that thing you built for us at Deliveroo?”

Only then did that uncomfortable idea — to create a product and start a company — pop into my head. When it became too loud to ignore, I took my savings and invested them in my own startup.

From Complaints to Opportunities

When I was a consultant, I kept finding the same bottlenecks. First you centralize CI/CD. Then you try Terraform Cloud/Terraform Enterprise, but you realize multiplayer mode isn’t really there.

And once teams hit a certain size, the problems stop being about “How do we write better HCL?” and become:

Who gets to run applies?

How does anyone audit what changed and why?

What happens when two people change the same thing?

How does compliance get answers without turning every deploy into a committee meeting?

Those gaps between good tools and workable processes kept showing up everywhere I went. Spacelift started as my attempt to close them.

Even then, I didn’t feel like a founder. And I still don’t. I feel like an engineer with a stubborn streak and a long list of complaints.

Which I’ll keep working to solve.

Thanks for reading,

Marcin

P.S. If stories like this resonate, join IaCConf on January 28 to hear practitioners unpack how AI is starting to shape the messy middle between “great tools” and “working systems” in real IaC environments.