Microagi raises $55 million seed round for industrial robot training data

Microagi GmbH has raised $55 million in seed funding to build a data and deployment layer for industrial robotics, the Munich startup said. Hummingbird led the round, with Northzone, LocalGlobe, Village Global and Redalpine participating; Microagi did not disclose its valuation, revenue or headcount.

The financing is large for a company that says it was founded about 10 months ago, and it reflects investor demand for robotics companies that can address a less glamorous bottleneck in physical AI: training data from real work environments. Microagi is not selling robots and says it is not developing its own robotics foundation models. Its pitch is that factories already buying robots and models still need a way to collect usable operational data, tune systems for specific tasks and improve performance after deployment.

The company was founded by engineers from Formula One teams Red Bull Racing and Mercedes-AMG Petronas. Microagi is led by Bercan Kilic, who previously worked as an aerodynamics engineer at Red Bull Racing in 2023 before leaving to start the company.

Atlas sits between robots, models and factory floors

Microagi’s core product is Atlas, which the company describes as a robotics data and deployment platform for industrial customers. According to Microagi, Atlas uses dedicated recording hardware and a secure ingestion system to capture and organize factory data, then uses that data to fine-tune existing robotics models for plant-specific work.

The company says the system continues collecting operational data after deployment, creating a feedback loop intended to improve accuracy over time. Chief Technology Officer Nico Nussbaum said Microagi’s engineers work on-site with customers after robots and models are already in place, using real production data to feed improvements into later runs.

That positioning is narrower than the usual humanoid robotics pitch. Microagi is selling infrastructure for training and deployment rather than a machine, which may make it easier to insert into existing industrial projects. It also means the company’s progress depends on customer access to factory data, model partners and robot deployments that Microagi does not fully control.

Microagi has not named customers. The company told Business Insider that five companies are collecting data through Atlas and that one customer is preparing to deploy robots in a factory. Those customers span automotive, logistics and food, according to Business Insider.

Consumer services are being used to collect robotics data

Microagi has also drawn attention for Shift, a consumer-facing data collection effort in New York City. Shift has offered free home cleaning by professional cleaners who wear body-mounted cameras, with the stated goal of recording first-person cleaning footage to train future household robots. Shift’s website says the cleanings are free for a limited time.

The tactic points to a broader problem for robotics AI. Text and image models have benefited from large public data sets, while robot learning requires video and movement data that show people performing tasks in ways machines can translate into action. The useful material is more constrained: clear views of hands and arms, consistent lighting and examples that can be converted into movement signals.

Microagi is also launching a private-chef service in San Francisco for the same data-collection purpose, according to the company. That effort is aimed at gathering data for robotic cooking.

The $55 million round gives Microagi capital to pursue both industrial deployments and data collection programs. The company has disclosed early customer activity, but not production-scale outcomes, commercial terms or whether its reinforcement loop produces measurable gains across factories. Those details will determine whether Atlas becomes a defensible layer in robotics or another AI services wrapper around scarce customer data.

This story draws on original reporting from SiliconANGLE.