Hi,

Something important is happening in DevOps right now, and it is bigger than another CI/CD trend.

For years, DevOps has been about automation: faster builds, faster tests, faster deployments, and shorter feedback loops. But the industry now appears to be moving into a new phase. The focus is shifting from automated pipelines to autonomous systems.

Over the past few months, major platform vendors have started moving in the same direction. AWS recently announced the general availability of its new DevOps Agent, describing it as an “always-available operations teammate” that can investigate incidents, correlate telemetry, improve reliability, and help prevent failures across cloud environments. You can read the AWS announcement here and the product page here.

InfoQ called the launch one of the first serious production-grade attempts at AI-powered incident investigation. At the same time, Forbes reported that AWS is building autonomous agents across the full software lifecycle, from coding and operations to security.

That matters because the DevOps problem is changing.

The old challenge was speed. Engineering teams wanted to ship faster, deploy more often, and reduce manual handoffs. But AI is now accelerating software creation itself. A recent Business Insider report noted that more companies are openly talking about how much of their production code is generated or co-authored by AI systems. GitLab’s latest Global DevSecOps Survey points to the same tension: AI is increasing development velocity, but it is also creating new operational bottlenecks, governance issues, and reliability concerns.

In other words, generating code faster is no longer the hard part.

The harder question is what happens after that code exists.

Who reviews it?
Who secures it?
Who deploys it?
Who monitors it?
Who understands the system when something breaks?

This is one reason platform engineering has become so important. As software delivery becomes more complex, more organizations are moving away from fragmented toolchains and toward internal developer platforms. These platforms standardize workflows, permissions, observability, deployment patterns, and compliance requirements.

PlatformEngineering.com recently described this shift as “the new DevOps standard.” Databricks has made a similar argument, framing internal developer platforms as a way to reduce cognitive load and help teams manage growing infrastructure complexity.

The emerging model is clear: developers focus more on intent, while the platform handles more of the orchestration, policy, compliance, and operational guardrails.

Security is evolving in the same direction.

The conversation is no longer just about DevSecOps. It is increasingly about AI governance. TechRadar recently warned that reusable “AI agent skills” could become a new enterprise supply-chain risk, especially as agents gain the ability to execute scripts, trigger workflows, and make infrastructure changes with elevated permissions.

GitLab is already moving into this territory with agentic remediation systems designed to automatically fix vulnerabilities and configure security pipelines. Academic research is also beginning to study how AI agents interact with CI/CD systems in real production repositories, including how AI-generated changes perform across GitHub Actions workflows.

Taken together, these developments point to a very different model of software operations.

The old DevOps model looked like this:

The new model is starting to look more like this:

That is a major shift.

The pipeline is no longer the center of the system. The platform is. And increasingly, the platform itself is becoming intelligent.

The teams that win over the next few years probably will not be the ones that simply ship the fastest. They will be the ones that can safely coordinate AI agents, developer workflows, operational policy, security controls, and infrastructure without losing control of the system underneath.

DevOps is not going away.

It is becoming more autonomous.