Machenta, a London-based artificial intelligence startup developing foundation models for robotics, raised $40 million in a seed funding round co-led by Creandum and Union Square Ventures. The funding was announced on October 27, 2026, as the company emerged from stealth.
Founded in 2025 by Professor Edward Johns and Vitalis Vosylius, Machenta is developing technology designed to enable robots to learn new physical tasks from a single demonstration without additional task-specific training.
Johns serves as co-founder and CEO and is a Professor of Robot Learning at Imperial College London. Vosylius serves as co-founder and CTO. The company builds on six years of research conducted at Imperial’s Robot Learning Lab, where the founders developed techniques for one-shot robot learning.
Machenta’s technology uses a foundation model trained entirely on simulated data rather than real-world training datasets. Its vision-demo-action approach processes a demonstration alongside visual observations, enabling a robot to generate actions for a new task without conventional post-training.
According to the company, the model can adapt to changes in object positions and task conditions after observing a single demonstration. Machenta has presented demonstrations involving packing, tidying, object manipulation and other physical tasks across different robotic platforms, including humanoid systems.
The startup’s research focuses on reducing the time and data required to teach robots new skills. Its approach uses simulation-generated training data to develop models capable of adapting to unfamiliar tasks through demonstrations.
Machenta plans to use the new capital to scale its robot learning model, expand computing capacity and grow its research and engineering team. The company is also expanding operations at its new office and laboratory in King’s Cross, London, as it continues developing the technology.
