Komodor launches agentic operations platform for SRE

Komodor has launched its Agentic Operations Platform for SRE, DevOps and platform teams.

The platform combines pre-built automation workflows with tools that let organisations build or import their own agents and manage them under a common set of controls.

Komodor is launching the product as software teams face pressure from a growing volume of changes moving into production. Coding agents are helping engineers ship faster, but that pace is also driving more incidents, support work, maintenance demands and spending.

The company cited a recent survey that found 60% of senior enterprise leaders are deploying agents in production. It also pointed to a Gartner forecast that more than 40% of agentic AI initiatives will be decommissioned by 2027 because of governance gaps, unclear returns or rising costs.

Ready-made workflows

The platform includes a set of workflows for operational tasks that teams often need to automate quickly. These are grouped into AI SRE, AI software operations and cost optimisation.

Across those categories, Komodor listed troubleshooting and incident management, alert intelligence, reliability work, cloud cost reduction, observability cost reduction, Kubernetes cost reduction, change intelligence, CI/CD health and production readiness as initial use cases.

The workflows include more than 50 specialist agents, integrations and related components that can be configured. Teams can add or remove steps, change routing and insert their own agents to fit internal processes.

Custom agents

Alongside the pre-built options, customers can create custom agents from scripts, runbooks or existing skills. They can also import agents built with third-party frameworks or develop new ones using Komodor's software development kit.

According to Komodor, those agents can run in on-premise environments or public cloud settings. Customers can also use different model providers while keeping agents tied to the same organisational context, memory and operating controls used by Komodor's own software.

The platform is also designed to test and compare agent versions before wider deployment. Teams can shadow-test changes, review performance against scenarios and production outcomes, and choose which model should handle a task before promoting updates.

Governance focus

Governance is a central part of the launch. Role-based policies determine who can invoke an agent and which credentials or tools it can access.

Komodor said guardrails check inputs, tool calls and model responses before an agent acts. Higher-risk actions can be held for human approval, while spending limits and audit trails are intended to help teams review activity and control costs.

The approach reflects a wider market concern over how companies move from AI pilots to production systems without losing oversight. Buyers are increasingly looking for ways to standardise controls as different teams experiment with their own agents, models and workflows.

The latest product builds on infrastructure already used in Komodor's AI SRE platform. Experience from that earlier work shaped the underlying context, memory and governance framework now being offered to customers that want to run their own agents.

Itiel Shwartz, Co-Founder and CTO of Komodor, said the main operational challenge is less about building an agent than managing it once deployed.

"The hardest part of running agentic operations in production is not building the agents. It is keeping every agent grounded in the right context, giving it memory that persists across incidents, ensuring that accuracy improves rather than degrades, and securing every run - all while keeping model and compute costs under control," said Shwartz.

He said the company had spent several years addressing those issues in its existing platform.

"We've spent five years solving those problems inside Komodor's AI SRE platform. The Komodor Agentic Operations Platform gives customers that same production-grade foundation from day one and a framework to build, run, and optimize their own agents leveraging the same context, memory and governance we built for ours," said Shwartz.

Komodor said organisations including Nebius, Dell, Cisco and Prudential have used its software. It also said it has raised USD $90 million in venture funding from investors in the US and EMEA.