How platform teams are using AI to keep up with rising software development velocity without sacrificing control
The speed at which software is developed has changed fundamentally. With AI-assisted coding now embedded in daily workflows, infrastructure requests arrive faster, earlier, and with less tolerance for waiting. Platform teams are expected to respond at the speed of experimentation while still maintaining security, compliance, and operational consistency.
That tension is not theoretical. We hear it directly from teams trying to balance self-service with control, and experimentation with governance. As we worked toward the general availability of Spacelift Intent and introduced Spacelift Intelligence, we spoke with some of our early access design partners to understand where AI genuinely helps platform engineers keep up with the demand for infrastructure, and where guardrails still matter most.
What follows are two real-world perspectives from platform leaders who explored Spacelift Intent as a way to bridge that gap.
When self-service becomes unmanageable
At Cityblock Health, developers already had access to self-service infrastructure. The challenge was what happened afterward.
Engineers could quickly spin up development environments, but much of the work beyond initial setup happened manually. Over time, this created a growing surface area of infrastructure that was difficult for the platform team to track, govern, or safely promote into staging and production.
As Logan Stuart, Director of Engineering at Cityblock Health, described it, the issue was not a lack of tooling but a mismatch between developer expertise and cloud complexity:
“Most software engineers know a lot about programming languages and very little about the cloud. Asking them to think in terms of Terraform modules, networking primitives, or routing is a big jump, and it often slows them down or blocks progress.”
This gap created two downstream problems. Developers either waited for platform support, or they worked around infrastructure best practices in the name of speed. Neither outcome scaled well.
Spacelift Intent introduced a different interaction model. Instead of forcing developers to understand cloud constructs up front, teams could describe what they needed in natural language while platform-defined policies are enforced behind the scenes. Intent provides a fast onramp for deployment to developers, but has built in audit trails, state management and drift detection, so Platform teams maintain full control of what Intent can deploy.
For Cityblock, that shift was less about replacing Infrastructure as Code and more about changing how teams reached it:
“The promise that stood out was letting developers express what they need without having to know every cloud detail, while still being able to turn that work into Terraform later and treat it as real infrastructure.”
From a platform perspective, this meant experimentation could happen earlier and more safely. Developers moved faster without creating untracked infrastructure, and platform teams retained a clear path back to governed, versioned code.
Creating safe spaces for experimentation
At Vega, the platform team approached Intent with a different but related goal: enabling experimentation without expanding risk.
Joe Hutchinson, Platform Lead at Vega, viewed Intent primarily as a prototyping tool, something that needed clear boundaries to be effective. From the start, isolation and guardrails were part of the design.
“Intent is a different way of provisioning infrastructure, and for us it has been best suited for experimentation. Running it in an isolated environment makes that clear and gives teams confidence to try things without worrying about unintended impact.”
That framing helped Vega create a safe on-ramp for engineers who were curious but not deeply experienced in infrastructure. By limiting where Intent could operate and pairing it with role-based access controls, the platform team could offer freedom without broad permissions.
The value, Joe noted, was not just speed, but clarity:
“It works well when people understand what it’s for. Once you frame it as a place to explore ideas quickly, without touching production systems, the workflow makes sense and adoption becomes much easier.”
For platform teams, this kind of intentional separation reduces friction. Engineers get faster feedback loops, and platform owners avoid the operational burden of cleaning up unmanaged experiments.
What changes when AI supports platform work
Across both experiences, a common theme emerged.
AI is most effective when it helps translate intent into action while respecting the rules platform teams already rely on.
Rather than replacing IaC or GitOps workflows, AI-assisted infrastructure works best as a complement. It lowers the barrier to entry, accelerates learning, and shortens feedback cycles, while allowing teams to promote Intent-created resources to Terraform and traditional GitOps pipelines when the Platform team wants to deploy through their existing operational process.
This is the philosophy behind Spacelift Intent and Spacelift Intelligence. Platform teams define the policies, constraints, and approved patterns once. Developers gain faster access to infrastructure through natural language, without bypassing security or compliance expectations.
As developer velocity continues to accelerate, the challenge is no longer whether platform teams should adopt AI, but how they do so responsibly. The experiences shared here point to a clear answer: pair AI-driven workflows with strong guardrails, and use them to amplify platform expertise rather than replace it.
That balance is what allows platform teams to keep pace without losing control.
Keep infrastructure moving at AI speed
Spacelift Intelligence keeps platform teams ahead. Fuse traditional IaC and GitOps pipelines with an AI deployment model and a powerful Infrastructure Assistant.
