Devtron Adds AI Agents to SRE Platform for Kubernetes Environments
Devtron today revealed it has added artificial intelligence (AI) agents to its open source platform for automating site reliability engineering (SRE) workflows across Kubernetes environments.
Announced at the Kubecon + CloudNativeCon North America 2025 event, version 2.0 of the Devtron platform also adds support for KubeVirt, an open source framework for encapsulating kernel-based virtual machines (KVMs) in a way that enables monolithic applications to run on Kubernetes clusters.
Devtron has also added financial operations (FinOps) tools to provide DevOps teams with the insights needed to better control Kubernetes costs.
Finally, Devtron has added support for graphical processor units (GPUs) that are increasingly being added to Kubernetes clusters running AI workloads.
Devtron CEO Ranjan Parthasarathy said these latest additions to the platform extend the ability of the Devtron platform to provide a single pane of glass for automating SRE workflows. The overall goal is to make it possible for SREs to autonomously maintain resilience and eliminate operational blind spots by relying on AI agents to automatically execute pre-approved runbooks.
It’s not clear how widely DevOps teams running Kubernetes clusters have adopted SRE automation platforms, but Devtron claims its platform is running at more than 21,000 installations spanning in excess of nine million deployments.
At the core of that platform are instances of the open source Argo continuous delivery (CD) platform, Helm, for deploying software on Kubernetes clusters and Flox, an open source framework for managing software and any associated dependencies that have been integrated via a single control plane to optimally run on Kubernetes clusters. The addition of the AI agents serves to then make the entire Devtron platform able to invoke natural language in a way that makes the platform more accessible to teams building and deploying cloud-native applications, said Parthasarathy.
It’s not clear to what degree organizations are migrating away from legacy DevOps platforms as they build and deploy applications on Kubernetes clusters, but many are clearly adopting best platform engineering practices to manage workloads at scale that are also becoming increasingly more complex.
The challenge, as always, is that the number of SREs that have Kubernetes expertise is limited. Devtron is making a case for using AI agents to enable fewer SREs to effectively manage a larger number of applications without necessarily having to expand the size of the DevOps team.
Human SREs will, of course, still be needed to validate the output generated by those AI agents but the expectation is that much of the toil those teams currently encounter will be significantly reduced, said Parthasarathy.
Each DevOps team will need to determine for themselves how much faith to place in AI agents, but, hopefully, as more manual tasks are eliminated the job of being an SRE will become more rewarding. SREs, in the meantime, would be well advised to start pushing the envelope of what AI is capable today on the assumption it will only get better.
In fact, IT professionals did not study to become SRE so they can babysit applications. SREs, instead, should be concentrating on how to design, for example, blueprints and runbooks that ensure applications are available and secure with, most importantly, the least amount of manual intervention required.
