Bercan Kilic got his dream job in 2023, designing aerodynamics for Red Bull Racing while it was winning everything. He found the engineering magnificent and the point of it thin.

His Munich startup, microagi, has now raised $55m, which it says is the largest seed round a German company has secured. Hummingbird led, with Northzone, LocalGlobe, Village Global, and redalpine participating. The valuation was not disclosed.

What microagi sells is narrower than the funding suggests, and more interesting for it. It does not build robots, and it does not build models.

It records workers using cameras and sensor-equipped gloves, then uses the footage to teach existing robotics models to do a specific job inside a specific customer’s factory.

That is unfashionable in a sector where the money has gone to hardware, from Walden Robotics’ $300m launch to All3’s legged construction robots. microagi sells the layer between somebody else’s robot and somebody else’s model.

Five companies are collecting data through the platform, Kilic said, and one is preparing to deploy robots on a line. Customers span automotive, logistics, and food. He declined to name the model partners.

“We provide the labs with data, they provide us with models, and then we layer on proprietary data to make our customers happy,” he said.

That arrangement exists because of a detour. microagi planned to do deployment only, until the team found that most robotics models were not good enough to start from.

Kilic’s analogy is training a new hire: an adult picks up a factory job in a week, a child may never get it, and existing robotics models are still children.

So it built shift, which is the part you may have encountered without knowing whose it was. It went viral this year offering free apartment cleanings in New York in exchange for filming the cleaners doing dishes, mopping, and folding laundry. This week it started offering free private chefs in San Francisco.

shift now operates in 15 countries and pays more than 20,000 people to record themselves performing physical tasks, selling the footage to the labs building robot brains. It competes with Scale AI, Turing, and micro1, all doing versions of the same thing.

The reason the business exists is a genuine asymmetry. Language models were trained on the internet. Robots have no internet, a gap the Berkeley roboticist Ken Goldberg puts at 100,000 years, and no amount of capital collapses it directly. Hence the cameras.

Kilic’s answer on whether he is filming people out of their own jobs is demographic rather than moral. Europe and the US are running out of workers, and China is automating regardless. It installed 295,000 factory robots in 2024, 54% of the world’s total, against 34,200 in the US.

Adapting existing models rather than training one is also a bet on those models arriving, which is the theory Nvidia has spent the year seeding. If it works, microagi has a market. If not, it has a great deal of footage of people mopping.

“If you run factories, the math is already on your desk,” he said. “Your most experienced people retire this decade, and their replacements were never born. Reshoring only works if the robots do.”

The numbers support the frame. The EU’s median age hit 44.9 in 2025, up from 39.6 two decades earlier, and the European Commission estimates the bloc could lose 18.8 million workers by 2050.

Hummingbird’s Firat Ileri, who led the round and previously backed Lovable, Kraken, and Etched, said he was struck by how rarely anyone at the Munich office seemed to leave it.

The money goes on compute, expanding shift’s network, and a US presence run from New York. microagi employs 37 people; shift about 75.

Kilic’s four co-founders include a former Mercedes F1 engineer, Yoan Iliev, and an ex-Alan Turing Institute researcher, Anton Poletaev.

Kilic thinks robotics is at its “GPT-2 moment”, with the scaling recipe not yet worked out but close. He also wants microagi to be the world’s largest company inside five years.

“In five years, if we haven’t deployed more than 20 million or 30 million robots, it’s a big failure,” he said. The company currently has one customer close to putting robots on a factory floor.