It was only a few years ago that perhaps the biggest hurdle for DevOps advocates was convincing leadership that it was worth the investment. That conversation has since shifted. In most organisations today, the value of faster releases, tighter feedback loops, and closer collaboration between development and operations is obvious. The challenge is no longer persuading people that DevOps matters but figuring out how to make it work in practice and, increasingly, figuring out how to make it work safely, as AI and automation open up both opportunities and fresh security concerns.
For many teams, the problem is not a lack of enthusiasm or ambition but a shortage of resources and skills. They want to automate more, streamline workflows, and adopt new practices, yet often find themselves already operating at full capacity just in keeping existing systems running. In that environment, the slightest of steps toward more advanced automation strategies can feel like a big leap forward.
This is where Internal Developer Platforms (IDPs) have started to prove their worth. By standardising infrastructure, consolidating tooling, and automating repetitive operational tasks, IDPs can take a significant amount of pressure off delivery teams, creating room for innovation without demanding extra headcount. However, while platforms and automation help ease the operational load, DevOps is as much about cultural transformation as it is about technology. Building trust between teams, changing long-established processes, and embedding new ways of working takes sustained effort, and that is often the harder part of the journey.
When Interest Outpaces Action: DevSecOps and AIOps
If DevOps itself has crossed into the mainstream, then DevSecOps and AIOps are in a different phase of the adoption curve. Arguably, the two most discussed topics in technology today are AI and cybersecurity (probably in that order) and the tone of conversations is usually somewhere between curiosity and cautious experimentation.
From an ‘Ops’ perspective, a lot is being said about both approaches, and it is becoming rare to meet a team that has not at least explored their possibilities. But there is still a noticeable gap between awareness and widespread implementation.
On the security side, the logic behind DevSecOps is compelling. More companies are realising that security has to be baked in from day one, not bolted on later. The difficulty lies in making that shift a practical reality, as integrating security checks early in the pipeline often requires new tooling, changes to established workflows, and in some cases, rethinking the roles and responsibilities within the team. These are not small adjustments, and when teams are already stretched, the prospect of introducing such changes can be daunting.
AIOps faces a similar dynamic. The concept of using machine learning to detect anomalies, predict incidents, and automate responses is attractive, but moving from early models to reliable, production-ready adoption takes more than a technology upgrade. It requires confidence in the data, trust in the recommendations generated, and clear processes for when human oversight is still necessary.
AI is Forcing the Security Issue
The rise of AI not only brings operational intelligence, it also introduces new risks. When organisations let AI loose on their data, they are also creating new attack surfaces and vulnerabilities. Sensitive information can be exposed through poorly governed models, whilst potentially adversarial inputs can manipulate predictions and automation that speeds delivery can also accelerate mistakes if security is not embedded from the outset. In other words, the more power you hand to AI-driven systems, the more critical it becomes to harden these systems against misuse.
This is why the conversation around AIOps cannot be separated from DevSecOps. As AI becomes more deeply woven into infrastructure and operations, teams must ensure that every layer of the pipeline is secure. The two trends are growing in tandem: DevSecOps provides the guardrails, while AIOps provides the intelligence. Together, they define what modern, resilient DevOps will look like.
There is definitely a momentum building. Cross-discipline Ops teams are beginning to run focused pilots that address specific operational pain points, and each small success helps make the broader vision feel less like a distant aspiration and more like an achievable next step.
Deployment That Gets Out of the Way
When it comes to deployment methods, the best platforms are those that allow developers to choose the right approach for the job without forcing them to navigate a maze of processes or context switches. In practice, that means supporting everything from traditional on-premise deployments to fully cloud-native, containerised workflows, but doing so through a consistent, accessible, and ideally self-service interface.
For many teams, a GitOps-first model has emerged as the most effective way to achieve that balance. By using version-controlled configurations to drive deployments, GitOps delivers a level of consistency and traceability that is hard to match, while also enabling rapid and safe rollouts.
But deployment success is not solely about what happens under the hood. It is also about the developer experience at the point of interaction. From a Cycloid perspective, that is why the introduction of Plug-ins has been such a significant boost for many of our customers.
Plug-ins make it possible to integrate key tools, surface metrics, and display dashboards directly inside the developer portal. That means that whether you are bringing in error tracking from Sentry, code quality checks from SonarQube, or custom logs tied to a specific environment, a developer can access the information they need in the context where they need it, while platform teams maintain oversight and ensure consistency across the ecosystem.
The Future of AI and ML Infrastructure Ownership
As AI and machine learning continue to move from experimental projects into core business functions, the question of who will be responsible for setting up and maintaining the infrastructure that supports these workloads is becoming increasingly relevant. Right now, there is no single answer, but patterns are beginning to emerge.
In many organisations, it is the existing DevOps or platform teams that are best positioned to take on this responsibility, extending their remit into what is often referred to as MLOps. These teams already have experience building and maintaining shared infrastructure, managing pipelines, and ensuring operational stability at scale, so expanding those capabilities to handle data science and machine learning workflows can feel like a natural evolution. This approach has the added benefit of keeping infrastructure governance consistent across application and ML projects, which reduces duplication of effort and makes integration smoother.
That said, as adoption grows, we can also expect to see more specialised MLOps roles appearing, particularly in larger enterprises or in organisations where AI is a major strategic focus. These specialists bring a deeper understanding of model lifecycle management, data handling requirements, and performance tuning for ML-specific workloads. However, rather than existing in isolation, they are likely to work closely with platform teams.
We are still in the early stages of this shift, but the potential for collaboration and shared ownership is significant, and it will be fascinating to see how these roles and responsibilities settle in the coming years.
DevOps is Evolving. How to Stay Ahead
The story of DevOps today is not one of selling the idea, but of scaling it. The principles are well understood, and the business case is rarely in question, yet the path to implementation is still shaped by the practical realities of resources, skills, and cultural readiness. Emerging practices like DevSecOps and AIOps show just how far the discipline can evolve, but they also reveal the gap that often exists between recognising a good idea and embedding it into daily workflows.
The most successful teams will be those that focus on creating a unified, self-service platform experience where automation, security, and visibility are not bolt-on features but fundamental building blocks. Whether that means standardising deployments through GitOps, integrating the right tools directly into developer portals, or extending infrastructure capabilities to support AI and ML workloads, the goal remains the same.
DevOps has always been about shortening the distance between an idea and its delivery to users, and that is not changing. What is changing is the scope of what we are delivering and the complexity of the systems involved. And as AI becomes more deeply embedded in both applications and operations, the organisations that succeed will be those that treat security as inseparable from innovation. By investing in both the cultural and technical foundations now, organisations can lay the rails that will carry not just their next release, but the next generation of their digital capabilities.
