The Evolution of Platform Engineering in the Age of AI: A New Frontier

As artificial intelligence continues to reshape the software development lifecycle, the role of platform engineering is undergoing a fundamental transformation. What was once a discipline focused primarily on managing infrastructure, CI/CD pipelines, and developer tooling has now expanded to include the orchestration of AI-driven workflows, the management of autonomous agents, and the creation of "paved paths" that operate at machine speed.

Recently, industry experts gathered for an InfoQ roundtable—moderated by Renato Losio—to discuss how AI is altering the landscape. The panel featured Davide de Paolis (Engineering Manager at Shine), Stephane Di Cesare (Platform Experience Lead at DKB), Camila Macedo (Software Engineer and Kubernetes specialist), and Stephen Cihak (Field CTO at Harness). Their consensus was clear: AI is not replacing platform engineers; it is elevating them into architects of a more complex, high-velocity, and automated future.

The Intersection of AI and Infrastructure: A New Operating Model

The primary shift in platform engineering is the move from human-centric to machine-centric operations. Historically, platform teams designed workflows for developers. Today, those same platforms are increasingly being consumed by autonomous AI agents.

The "Neglected" Backlog

One of the most immediate benefits of AI in platform engineering is the remediation of technical debt. Davide de Paolis noted that platform teams are often overwhelmed by "important but never urgent" tasks—documentation, boilerplate code generation, and recurring infrastructure-as-code (IaC) updates. "Everything that was neglected is now accomplished," de Paolis explained, highlighting how AI allows teams to easily automate documentation and diagramming, tasks that previously suffered from a lack of human bandwidth.

Machine-Speed Deployment

Stephen Cihak emphasized that as AI reduces the time required to write code, the traditional bottlenecks in the software delivery process have shifted. "The bottlenecks now are in the outer loop—the deployment loop, the security scans, and the governance pieces," Cihak said. Because AI can generate code at unprecedented speeds, platform teams are under immense pressure to move away from manual phase-gates toward a fully autonomous Software Development Life Cycle (SDLC).

Chronology of the Shift: From DevOps to Autonomous SDLC

The evolution of these systems has been rapid, moving through distinct phases over the last two decades:

• The Era of Manual Ops (Pre-2010s): Infrastructure was largely managed manually or through basic scripting. Developers were disconnected from the deployment process.
• The DevOps Revolution (2010s): The emergence of cloud-native technologies, Kubernetes, and IaC (Terraform) allowed for automation. This introduced the concept of the "DevOps engineer" who bridge the gap between development and operations.
• The Platform Engineering Era (2020–2023): Organizations began centralizing tools into Internal Developer Platforms (IDPs), such as Spotify’s Backstage, to improve developer experience and reduce cognitive load.
• The AI/Agentic Era (2024–Present): Platform teams are now building "guardrail-as-code" frameworks to support AI agents that generate infrastructure, fix vulnerabilities, and handle incident response at machine speed.

Supporting Data: The Productivity vs. Governance Paradox

While AI-assisted coding is drastically increasing developer output, it presents significant risks to stability and security. Data from current industry trends suggests that while AI can improve productivity by 20–40% in coding tasks, it also increases the risk of "configuration drift" and security vulnerabilities.

Stephane Di Cesare pointed out that in the banking sector, the "bottleneck is not the coding itself, but compliance and security." Consequently, the platform team’s job has become more about centralizing compliance logic so that AI agents can "consume" these policies during the build process, ensuring that speed does not come at the expense of regulatory failure.

Official Perspectives: The Role of the Platform Team

The panelists shared distinct perspectives on how the responsibilities of platform teams are evolving in response to the AI wave.

On Guardrails and Governance

The consensus among the experts is that AI does not remove the need for governance; it makes it mandatory. Stephen Cihak argued that because models are non-deterministic, platform teams must build "stronger, larger structures" to allow for higher speeds. "If you don’t have that, it’s like driving through the Outback—you’re going to hit bumps and rolls and you’re not going to be able to go as fast," Cihak warned.

On the "Human-in-the-Loop" Necessity

Camila Macedo offered a cautionary note regarding the autonomy of agents. "I have a lot of concerns about agents being free to take decisions without a human being," she stated. She noted that current benchmarks for code generation accuracy—often ranging between 25% and 65%—are insufficient for critical banking infrastructure. The platform, therefore, must remain the "source of truth" that mandates human oversight for high-impact changes.

On Cognitive Load and Metrics

Measuring the success of an IDP remains a challenge. Davide de Paolis noted that simple metrics like "number of tokens used" are irrelevant. Instead, platform teams should focus on:

1. Time-to-Production: Measuring the reduction in time from "idea to production" after the implementation of a new platform feature.
2. Onboarding Velocity: How quickly a new developer can be productive within the ecosystem.
3. Communication Friction: Monitoring the volume of Slack/Jira traffic regarding infrastructure queries; a decrease suggests that the platform’s self-service documentation is effective.

Implications for the Future

The Rise of the Autonomous SDLC

In the next five years, the industry is expected to move toward an "autonomous SDLC." In this model, a developer will define an intent, and a system of agents will handle the provisioning, coding, testing, and deployment. The platform team’s primary value will not be in building the pipelines themselves, but in maintaining the "knowledge graph" and the "guardrail engine" that prevents these agents from acting irresponsibly.

Identity Crisis and Professional Evolution

The experts rejected the notion that platform engineering roles will disappear. Stephane Di Cesare noted that AI is a tool, not a replacement. Just as compilers and high-level languages did not eliminate the need for programmers, AI will not eliminate the need for infrastructure experts. Instead, engineers will transition from "builders of plumbing" to "architects of automated systems."

The Return to Documentation

Perhaps the most ironic implication of the AI era is the renewed focus on documentation. For years, documentation was the most neglected aspect of software engineering. Now, because AI agents require structured context to function accurately, organizations are finally investing in high-quality knowledge bases, internal APIs, and standardized code conventions.

Conclusion: Engineering for the Long Term

As the roundtable concluded, the experts reached a unified perspective: AI is an evolutionary layer of abstraction. While the "shiny tools" of the week may distract some, the fundamentals of platform engineering remain the same: providing a stable, secure, and developer-friendly environment.

The successful platform teams of the future will be those that embrace AI as an engine for automation while doubling down on the "boring" but essential tasks of governance, security, and psychological safety. As Stephane Di Cesare aptly put it, "You don’t drive a car by bumping into the guardrails," but having those guardrails in place is exactly what allows the driver to accelerate with confidence. For the platform engineer, the mission remains constant: enabling others to build, while ensuring that what they build can withstand the test of time.