OpsMill raises €12 million to secure AI agents operating in critical infrastructures

The automation of IT infrastructure is entering a new phase. After several years focused on orchestration tools and DevOps pipelines, the rise of AI agents is gradually shifting the problem toward a far less visible, yet increasingly critical layer: the quality of the data describing the infrastructures themselves. It is on this front that OpsMill has announced a €12 million Series A funding round led by IRIS Capital, with participation from BGV as well as existing investors Serena and Partech.

The Paris-based company develops Infrahub, a platform designed to centralize, structure, and improve the reliability of infrastructure data used by network, cloud, and operations teams. The objective is not simply to better document IT environments, but to make this data reliably usable by automated systems and, increasingly, by AI agents capable of acting directly on critical infrastructures.

The issue is becoming strategic as companies accelerate the automation of their operations. In most large enterprises, infrastructure data remains scattered across legacy CMDBs, internally developed scripts, spreadsheets, or partially synchronized cloud tools. These fragmented environments were already difficult to maintain within traditional architectures. They become significantly riskier when automated systems are making decisions or executing changes at high speed.

The challenge is no longer limited to operational efficiency, as a misconfiguration propagated automatically can now trigger large-scale outages, compliance failures, or cybersecurity incidents. According to figures cited by OpsMill, the average cost of an infrastructure outage reaches approximately $300,000 per hour, or nearly €255,000.

In this context, OpsMill is attempting to build what the company describes as a “source of truth” for IT infrastructures. Unlike traditional tools that treat infrastructure components as static inventories, Infrahub is built on a graph-oriented architecture capable of modeling complex dependencies between servers, network equipment, cloud environments, security policies, and application services.

This approach makes it possible to contextualize relationships between technical components and govern modifications before they are deployed into production. Every change goes through a validation process inspired by DevOps practices, applied not only to application code, but directly to the data describing the infrastructure itself.

“Automation is fundamentally a data problem: if you only have a partial view of your network, you are operating blind. Writing the code that automates infrastructure has never been the real challenge — the real difficulty has always been maintaining it and being able to trust it in production. We built Infrahub so that infrastructure teams, and the AI agents working alongside them, always have access to a complete and reliable reference of what exists, what should exist, and a safe way to evolve it at scale,” explains Damien Garros.

In financial services, industrial environments, and telecommunications, where technical architectures are becoming increasingly hybrid and distributed, the ability to maintain a coherent representation of infrastructure is becoming a major operational challenge.

The company is also relying on an open-source strategy to accelerate adoption. Infrahub is available in a free Community edition and an Enterprise version commercialized under license, following a model popularized by GitLab and Elastic. The approach is designed to encourage adoption by technical teams before upselling governance, compliance, and large-scale operational capabilities.

OpsMill says its platform is already being used by several international organizations, including TikTok and Eurofiber, the European cloud and network services operator. According to the company, Eurofiber reduced certain deployment timelines from several days to just a few minutes after adopting the platform.