OpsMill raises $14m Series A to make IT infrastructure data trustworthy enough for AI agents

IRIS led the Paris- and London-based startup’s round, with BGV joining alongside existing investors Serena and Partech. The company’s Infrahub platform is in production at TikTok and at one European cloud provider that says it has cut deployment times from five days to fifteen minutes.

OpsMill, the Paris-headquartered infrastructure data management company, has raised $14m in a Series A round led by IRIS, with participation from BGV and existing investors Serena and Partech.

The company says the funding will go to growing its engineering and product teams and continuing to develop its Infrahub platform, which is designed to give AI agents and engineering teams a single trusted view of an enterprise’s IT infrastructure.

The pitch addresses a problem most enterprises have lived with quietly for years. While automation has spread through applications and workflows, the data describing the underlying infrastructure, physical hardware, virtual machines, cloud resources, and the connections between them, has remained scattered across spreadsheets, configuration management databases, and ad-hoc scripts.

None of those sources was designed to feed AI agents reliable information. When agents act on incomplete or inaccurate infrastructure data, the company says, the result is errors that can cascade through production systems quickly.

OpsMill cites two figures to size the problem. Gartner forecasts that 30 per cent of enterprises will automate more than half of their network activities by 2026, up from under 10 per cent in mid-2023. Per the ITIC 2024 Hourly Cost of Downtime Report, the average enterprise loses around $300,000 for every hour of downtime, before reputational costs are counted.

Infrahub, OpsMill’s flagship product, takes a different approach to representing infrastructure data than the table-based asset registers most enterprises use. The platform is built on a graph database that maps the connections between hundreds of thousands of infrastructure elements, including the metadata describing how each element should be configured.

Every proposed change is validated and approved through a DevOps-style review process before deployment. Co-founder and CEO Damien Garros, who built and scaled Infrahub alongside co-founder and COO Karen Gallantry, spent two decades on the operator side of the same problem at Juniper, Roblox, and Network to Code before starting OpsMill.

Garros put the case for the company in his own framing. “Automation is ultimately a data problem and if you only have a partial view of your network, you’re flying blind,” he said in the announcement.

“Writing the code for automating infrastructure was never the problem; the challenge 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 a complete, trusted record of what exists, what’s supposed to exist, and a way to safely change and evolve at scale.”

Infrahub is available in two editions, free open-source Community and licensed Enterprise. OpsMill compares the model to GitLab’s, with users able to develop on the open-source version and graduate to the Enterprise edition when they need governance and compliance features at scale.

The open-source community already includes hyperscalers including TikTok, alongside global retailers, fintechs, insurers, and manufacturers using the Enterprise edition in Europe and North America.

Eurofiber, a European cloud-services provider named in the announcement as an Enterprise customer, has cut its service deployment times from five days to fifteen minutes since deploying Infrahub. That figure is the most concrete customer-side data point in the company’s release.

Julien-David Nitlech, managing partner at IRIS, framed his fund’s investment in similar terms in the announcement.

“The race to adopt AI in enterprise infrastructure is real, but most organisations are trying to build on foundations that were never designed for it,” he said.

“OpsMill is solving the problem that everyone else is working around: without clean, structured, trustworthy infrastructure data, AI-driven operations simply cannot function at scale.”

OpsMill, headquartered in Paris with a London office, says the Series A will fund engineering and product expansion alongside continued development of data-centric AIOps capabilities.