# Machenta Raises a $40 Million Seed Round for a Robot Model Trained Only in Simulation to Learn Tasks From One Demonstration

Times of AI Desk · 2026-10-28 · Research

[https://timesof.ai/2026/10/machenta-40m-seed-one-shot-robot-learning-sim](https://timesof.ai/2026/10/machenta-40m-seed-one-shot-robot-learning-sim)

> Machenta, a London robotics start-up spun out of Imperial College London, has raised a $40 million seed round co-led by Creandum and Union Square Ventures, Business Insider and Imperial reported on October 27. The company says its model is pre-trained entirely on simulated data and can direct a robot to do a new task after seeing it performed once. Business Insider reports that the 11-person company has no near-term plans to commercialize its technology and that more precise tasks remain challenging.

Machenta is betting that robots can learn physical tasks without the large real-world datasets its better-funded rivals are collecting. Investors have put $40 million behind that approach, Business Insider reports, and the founders' published research shows both what it can do and where it has fallen short.

## The seed round

Machenta raised a $40 million seed round co-led by Creandum and Union Square Ventures, the company exclusively told Business Insider, and declined to disclose its valuation. Imperial College London, where the company's founders did the underlying research, calls it the largest ever seed raise by a UK robotics startup. Imperial says the money will scale Machenta's model and expand its team at a new office and lab in King's Cross, London.

## Learning from one demonstration

CEO and co-founder Edward Johns, an Imperial professor, told Business Insider that the company trains its model in computer-generated environments that mimic how objects move and interact, without using a "single pixel of real-world data." Once trained, Johns said, the model can direct a robot to do a new task, such as tidying a table or putting objects into a box, after watching it done once and without further training. Imperial says the team has shown a single model learning diverse tasks such as packing and tidying across any robot, including humanoid robots. Johns and former PhD student Vitalis Vosylius founded Machenta late last year after six years of work in Johns's Imperial lab, according to Business Insider.

## What the founders' research shows

Johns and Vosylius have published the idea before. Their paper "Instant Policy," posted to arXiv in November 2024, describes a model that learns a new task from one or two demonstrations without further training, trained on what the authors call pseudo-demonstrations: arbitrary robot trajectories generated in simulation. In the paper's real-world tests on 16 everyday tasks, 10 attempts each, it averaged 88.75% success, against 34% to 41% for the three baseline methods the authors adapted. The paper lists its own limits, including that the method "lacks the precision needed for tasks with extremely low tolerances or rich contact dynamics."

Machenta has not said whether its model is built on Instant Policy. The paper's precision limit matches what Business Insider reports about the company's model today.

## Early-stage, against well-funded rivals

Business Insider reports that Machenta has 11 employees, remains focused on research and has no near-term plans to commercialize its technology. Its model can learn some tasks immediately, the publication says, but more precise tasks and complex movements remain challenging. Business Insider names Physical Intelligence and FieldAI as California-based competitors, the latter [raising at a $10 billion valuation](https://timesof.ai/2026/10/fieldai-700m-10b-term-sheet). It adds that Skild AI unveiled a model in August that it said could perform previously unseen tasks lasting up to 10 minutes after a single video demonstration.

## Limits

- The capability claims come from Machenta and Imperial, including a video the company shared with Business Insider. No independent benchmark of the company's model has been published, and the Instant Policy results are the authors' own.
- The UK seed-round record is Imperial's and Machenta's claim.
- Business Insider puts the headcount at 11; Tech.eu says the team numbers around a dozen.
- No valuation was disclosed.

## Sources

- [Business Insider: An 11-person startup raised $40 million to challenge robotics' data-hungry giants (October 27, 2026)](https://www.businessinsider.com/this-11-person-startup-raised-40-million-train-robots-virtual-worlds-2026-10)
- [Imperial College London: Imperial AI spinout Machenta raises $40 million to make robots instant learners (October 27, 2026)](https://www.imperial.ac.uk/news/articles/2026/imperial-ai-spinout-machenta-raises-40-million-to-make-robots-instant-learners/)
- [arXiv: Instant Policy: In-Context Imitation Learning via Graph Diffusion, Vosylius and Johns (November 2024)](https://arxiv.org/abs/2411.12633)
- [Tech.eu: London robotics startup Machenta raises $40M seed investment (October 27, 2026)](https://tech.eu/2026/10/27/london-robotics-startup-machenta-raises-40m-seed-investment/)