Mowito Raises $3 Million to Teach Factory Robot Arms by Showing, Not Programming

Bengaluru- and Detroit-based Mowito has raised $3 million in pre-seed funding led by Version One Ventures, with PyTorch creator Soumith Chintala among its angels, to scale foundation AI models that let industrial robot arms learn tasks by demonstration. Its software already runs on production lines at a Fortune 500 automaker and a major electronics manufacturer.

July 9, 2026
4 min read
M

Manik Gupta

Founder and editor of DeepTech India. Manik writes about India's frontier technology ecosystem — AI, semiconductors, space, quantum, robotics and biotech — translating research and policy into clear, reliable reporting.

Mowito Raises $3 Million to Teach Factory Robot Arms by Showing, Not Programming
Newskart

Mowito, a robotics startup with roots in Bengaluru and a foothold in Detroit, has raised $3 million in a pre-seed round to expand the artificial-intelligence software that runs on industrial robot arms. The round, announced on 7 July 2026, was led by Version One Ventures, with participation from All In Capital, iSeed, Unisol and a clutch of angel investors — most notably Soumith Chintala, the creator of PyTorch, the deep-learning framework that underpins much of the world's AI research and production.

The thesis is simple to state and hard to build: teach a factory robot the way you would teach a person — by showing it — rather than by hand-coding every motion.

The problem with programming robots

Industrial robot arms are everywhere in modern manufacturing, but they are notoriously rigid. Traditionally, each task — a weld here, a part placed there — is painstakingly programmed, and any change to the product or the line means an engineer reprogramming the robot. That upfront effort is why automation has stayed concentrated in high-volume, slow-changing production and largely bypassed the messier, more variable work that fills most factories.

Mowito's answer is a foundation AI model for physical tasks. Instead of scripting motions, an operator demonstrates a task and the model learns to reproduce and generalise it, covering assembly, pick-and-place, inspection and logistics. The pitch is that a robot arm becomes something closer to a trainable worker than a fixed machine — reprogrammable by demonstration in hours rather than reprogrammed by code over weeks.

Physical AI, and why investors are paying attention

The label the industry has settled on is "physical AI": models that do not just generate text or images but act in the world through machines. It is one of the hotter frontiers in robotics, and the calibre of Mowito's backers reflects that. Chintala's involvement is more than a marquee name; as the person behind PyTorch, his bet signals that the technical approach is credible to someone who has seen most of the field's plumbing up close.

Founded in 2024, Mowito says its software is already running on production lines at a Fortune 500 automotive company and at one of the world's largest electronics manufacturers — unusually concrete traction for a company raising a pre-seed round, and a sign that the product has moved past the demo stage into paying, high-stakes environments where a mistake stops a line.

An India-built model, a US market

Mowito's structure — engineering talent in Bengaluru, commercial presence in Detroit, the historic heart of American automotive manufacturing — mirrors a pattern among ambitious Indian deep-tech firms: build the hard technology at home, sell it where the largest industrial buyers are. The startup says the fresh capital will go toward accelerating its expansion in the United States, strengthening its engineering and go-to-market teams, and scaling deployments across automotive and electronics manufacturers.

That places Mowito in a small but growing cohort of Indian robotics and physical-AI companies attacking the automation of real-world manufacturing rather than the software layer above it. The wider funding data for 2026 shows physical AI and robotics emerging as a distinct theme alongside sovereign language models and defence AI — deep tech that makes or moves physical things.

What it has to prove

A pre-seed round buys runway, not a moat. The claims that matter now are whether Mowito's model genuinely generalises across tasks and factories — the difference between a clever demo and a platform — and whether it can hold its own against far better-funded global robot-learning efforts. Early deployment at two industrial giants is a strong opening card. Turning that into a repeatable, defensible product is the work the $3 million is meant to fund.

Sources

Tags

MowitoVersion One VenturesSoumith ChintalaAll In CapitalPhysical AI