CVAT lands $2M from OH.io to commercialise its open-source AI labeling tool
What's the deal? CVATDealroom has a profile for this one. Try Dealroom →, an open-source platform for labeling image, video, audio, and 3D point cloud data used to train AI systems, has raised $2 million led by OH.ioDealroom has a profile for this one. Try Dealroom → Capital. The deal also adds CVAT to the portfolio of sister company OH.io, which pairs the capital with dedicated sales and marketing support aimed at North America.
What's the endgame? CVAT began as an internal Intel project in 2017 and became an independent company in 2022. Its free, self-hosted Community product has drawn more than 17,000 GitHub stars and over 3 million downloads; the funding targets converting that adoption into paying enterprise use of its cloud version, CVAT Online.
The market expansion: OH.io will deploy a Columbus, Ohio-based team to run marketing, sales, and community work, and grow CVAT's presence across North America. Initial target markets include manufacturing quality assurance, machine vision, robotics, geospatial AI, and precision agriculture.
Why now? CVAT is already used in robotic surgery training at the University of Illinois Chicago, in automotive driver-assistance systems, and by utilities inspecting infrastructure via drone imagery. OH.io, founded by Veeam Software co-founder Ratmir Timashev, embeds sales and marketing staff into AI startups.
"CVAT has grown into one of the most widely used tools of its kind largely through word of mouth," said co-founder and chief executive officer Nikita Manovich. The OH.io partnership, he said, provides "the sales and marketing muscle to reach companies that don't yet know how our technology can help them."
What could go wrong? The $2 million is modest next to rivals Labelbox, Encord, and Roboflow, which have raised far more. CVAT's growth will hinge on winning enterprise customers without weakening the open-source appeal that built its community.
The signal: "The next wave of AI won't be limited to screens," said Timashev. The deal is less a large financing round than a bet that the next phase of AI — in robots, vehicles, and factories that must "learn to see" — will reward tools that can label the physical world, if their open-source roots can be turned into revenue.
Read more: pulse2.com
Image credit: Generated with Gemini
