Puru Rastogi Mowito: The Founder Teaching Robots to Learn Like Humans

Puru Rastogi Mowito is the story of an engineer who spent a decade building autonomous machines before deciding factory robots were being taught the wrong way. Rastogi is the co-founder and CEO of Mowito, a Bengaluru- and Detroit-based startup building “physical AI” models that let industrial robot arms learn new tasks by watching a single human demonstration, instead of being hand-programmed line by line. In July 2026, Mowito closed a $3 million pre-seed round led by Version One Ventures, with backing from angel investors who rarely write early checks together, and with robots already at Denso and one of the world’s largest
electronics contract manufacturers. Here is how Rastogi got there, what Mowito actually does, and why investors are betting the next wave of automation depends on software, not steel.

Table of Contents

1. Puru Rastogi Mowito: Who Is He?
2. From Autonomous Helicopters to Robotic Trash Sorting
3. The Problem Mowito Set Out to Solve
4. Building Mowito: Team, Product, and NeuralPick
5. Inside Mowito’s $3 Million Pre-Seed Round
6. Customers and Early Traction: Denso and Beyond
7. The Advisors and Investors Backing Mowito
8. Mowito’s Position in the Physical AI Race
9. Leadership Style and What’s Next for Mowito
10. Lessons for Entrepreneurs From Puru Rastogi’s Journey
11. Frequently Asked Questions
12. Conclusion

Puru Rastogi Mowito: Who Is He?

Puru Rastogi is an Indian roboticist and entrepreneur who studied at Carnegie Mellon University and the Indian Institute of Technology, Guwahati, before spending several years building autonomous systems for other companies. According to his Crunchbase profile, he worked as a robotics engineer at Near-Earth Autonomy and served as co-inventor at CleanRobotics, a company known for its AI-powered trash-sorting systems, before founding Mowito. His personal website describes him simply as “a roboticist and a maker” who likes “creating autonomous machines,” a description that undersells a founder now building software that runs on real factory floors on two continents.

Rastogi splits his time between Bengaluru, San Francisco, and Detroit, a geography that mirrors Mowito’s own footprint: engineering in India, customers in the American Midwest’s automotive corridor. Public profiles also list a connection to University College Dublin, suggesting research ties beyond his two primary alma maters, though the bulk of his public robotics record traces back to CMU and IIT Guwahati.

From Autonomous Helicopters to Robotic Trash Sorting

Before Mowito, Rastogi’s engineering path ran through some of robotics’ hardest, least glamorous problems. According to Mowito’s lead investor, Version One Ventures, Rastogi built autonomous helicopters while studying at Carnegie Mellon, one of the toughest domains in robotics because there is no margin for software error mid-flight. He later worked at Near-Earth Autonomy, a CMU spinout focused on autonomous flight systems, before co-inventing the sorting technology behind CleanRobotics, a startup that uses computer vision to automatically separate recyclables from trash.

That combination, perception-heavy robotics for objects that vary wildly in shape, size, and condition, turned out to be direct preparation for Mowito. Sorting a stream of unpredictable trash and picking irregular parts off a factory conveyor belt are, from a software standpoint, close cousins. Rastogi has described this connective thread himself, telling one interviewer that his time at CleanRobotics gave him “insights into applying AI for autonomous tasks,” while his work at Near-Earth Autonomy “deepened his understanding of AI and robotics” in ways that shaped Mowito’s eventual approach to automation.

The Problem Mowito Set Out to Solve

Industrial robot arms have automated factories for decades, but only for tasks that never change. A traditional robotic arm needs to be explicitly programmed, motion by motion, for every part it touches. The moment a manufacturer swaps a component, changes a supplier, or shifts a production line, engineers have to reprogram the robot from scratch, a process that can take days and requires specialized staff most factories don’t have on hand.

Rastogi’s pitch is that this is now a software problem, not a hardware one. In a July 2026 statement announcing Mowito’s funding, he said manufacturing has reached a point where hardware is no longer the constraint, and that factory robots should learn the way people do, by observing and repeating a task rather than being reprogrammed every time something changes. Version One Ventures frames the market opportunity around that gap directly, estimating that rigid, hand-coded robot programming leaves roughly $110 billion of automatable work untouched in the automotive sector alone, because the cost and time of reprogramming makes automation impractical for anything but the highest-volume, least-variable tasks.

Building Mowito: Team, Product, and NeuralPick

Mowito’s core technology is a set of physical AI models that let standard, off-the-shelf robot arms learn a new task from as few as one human demonstration, without any hardware modification. The company calls its underlying system NeuralPick, an AI engine that combines computer vision and force feedback to let robotic arms pick, place, assemble, and inspect parts with reported precision down to roughly 200 microns, while adapting in real time to items that move, shift, or vary from unit to unit.

According to Mowito’s own site, the platform is built around three ideas: eliminating the jigs and fixed setups that traditional automation requires, giving factory-floor operators, not outside engineers, a no-code web interface to reconfigure the system themselves, and compressing what used to take months of integration work into a matter of days. The company’s product line spans vision-guided assembly for delicate, fixture-free parts; automated machine loading and unloading; and camera-based inspection that Mowito says can cut inspection costs by more than 60% by catching defects and missing parts on moving lines.

Rastogi doesn’t build Mowito alone, and the Puru Rastogi Mowito partnership only works because the founding team splits the work cleanly. Company materials describe a three-person founding team: Rastogi as CEO, Safar V as CTO, a graduate of IIT Madras, and Adityanag Nagesh as Chief Business Officer, described by Version One Ventures as a three-time founder. That mix, deep robotics research paired with repeat startup experience, is part of what convinced early backers the team could execute on a genuinely hard technical problem while still moving fast commercially.

Inside Mowito’s $3 Million Pre-Seed Round

Mowito announced a $3 million pre-seed round on July 7, 2026, led by Version One Ventures, with participation from All In Capital, Unisol, and iSeed. The round also drew a notable slate of angel investors: Soumith Chintala, CTO of Thinking Machines Lab and co-creator of the PyTorch framework used across the AI industry; Adarsh Kulkarni of Foundry Robotics; Ashish Kulkarni of Coformer.ai; and Vaibhav Domkundwar of Better Capital.

The company says the capital will fund its expansion into the United States, alongside growth of its engineering and go-to-market teams as it scales deployments with automotive and electronics manufacturers. Kushal Bhagia, a partner at All In Capital, pointed to Mowito’s technical depth and early customer validation as central to the firm’s decision to invest, a combination that is unusually concrete for a company still at the pre-seed stage.

Customers and Early Traction: Denso and Beyond

What separates Puru Rastogi Mowito from many pre-seed robotics startups is that its technology is not confined to a lab. According to Mowito’s own materials, the company has deployed robotics solutions inside factories operated by Denso, the Japanese automotive components giant, and a Fortune 500 electronics manufacturing services (EMS) provider. Founders and investors alike describe the deployments as running on live production lines, handling tasks such as fastening, wire-harness assembly, and pick-and-place work, rather than pilot demonstrations.

That distinction matters in a robotics funding market crowded with impressive demos that never reach a real factory floor. Mowito also recently won recognition at a 2026 industry robotics challenge focused on adaptive automation for wiring-harness manufacturing, a notoriously fiddly task given how much harness assemblies vary between vehicle models. For a startup barely two years old, having named, large-scale manufacturing customers already running its software in production is a meaningfully strong traction signal.

The Advisors and Investors Backing Mowito

Beyond its institutional investors, Mowito has assembled an advisory bench that Version One Ventures has called among the strongest in Indian physical AI. Soumith Chintala, who co-created PyTorch and now serves as CTO of Thinking Machines Lab, advises the company alongside Lerrel Pinto, a robotics researcher recognized on MIT Technology Review’s TR35 list and affiliated with Meta’s Systems and Interaction Lab.

That advisory bench sits alongside a broader wave of researcher-backed AI startups, from Fei-Fei Li’s World Labs to Mowito, where credibility in the research community increasingly doubles as an investment signal. Boris Wertz, founding partner at Version One Ventures, has framed the investment as consistent with his firm’s broader robotics thesis: that task-specific robots built to do one job exceptionally well will outcompete generalized humanoid robots built in search of a problem to solve. Mowito, in his view, fits that mold precisely, a system engineered for the narrow, high-precision demands of assembly work, already proving itself on real production lines rather than in a demo reel.

Mowito’s Position in the Physical AI Race

Mowito operates inside one of 2026’s most heavily funded robotics categories, often described as “physical AI,” where startups apply large-model techniques to teach robots to interact with the real world, much as enterprise AI infrastructure players like Runlayer are doing for digital systems. Competitors and adjacent players range from humanoid-robot component suppliers building dexterous hands, to companies developing general-purpose robot foundation models for warehouses and logistics.

Mowito’s differentiation is narrower and more pragmatic: rather than chasing general-purpose humanoid robots, it focuses squarely on making existing, commodity industrial arms smarter through software, with no new hardware required. That approach lowers the barrier for factories to adopt the technology, since manufacturers can deploy Mowito’s models on equipment they already own, rather than waiting for new robotic hardware to mature and come down in cost.

Leadership Style and What’s Next for Mowito

Rastogi has described Mowito’s approach to customer relationships and product design as deliberately hands-on. In past interviews, he has emphasized giving factory-floor operators direct control over reconfiguring the system themselves, rather than requiring specialized engineers for every change, a philosophy that shows up directly in Mowito’s no-code interface. That operator-first instinct appears to extend to how Rastogi talks about the company’s growth: the new funding is earmarked specifically for U.S. expansion and scaling deployments with existing automotive and electronics customers, rather than a pivot into new verticals.

With teams already split between Bengaluru and Detroit, and named production customers in both the automotive and electronics supply chains, Mowito’s near-term roadmap looks less like a moonshot and more like disciplined, customer-led scaling, expanding the number of tasks its models can learn and the number of factories willing to hand a robot a new job after watching a human do it once.

Lessons for Entrepreneurs From Puru Rastogi’s Journey

Rastogi’s path offers a few instructive patterns for founders building in deep tech, echoing lessons from other young technical founders like Surya Midha at Mercor. First, his years at Near-Earth Autonomy and CleanRobotics weren’t a detour before Mowito; they were direct technical preparation, teaching him how to build perception systems for unpredictable, real-world objects long before he applied that skill to factory parts. Second, Mowito’s insistence on running its models on existing, off-the-shelf hardware rather than building new robots from scratch let the company reach live production deployments with a tiny founding team and a modest first check. Third, building a technical advisory board of genuinely credible names, not just recognisable ones, gave early investors concrete signals of quality that a pre-seed pitch deck alone rarely provides.

Frequently Asked Questions

Who is Puru Rastogi? Puru Rastogi Mowito refers to the founder-and-company pairing behind one of 2026’s more closely watched physical AI startups. Rastogi is the co-founder and CEO of Mowito, building software that teaches industrial robot arms to learn new tasks from human demonstration. He previously worked at Near-Earth Autonomy and CleanRobotics after studying at Carnegie Mellon University and IIT Guwahati.

What does Mowito do? Mowito builds AI models that let standard robot arms learn manufacturing tasks, such as assembly, loading and unloading, and inspection, from as few as one human demonstration, without requiring new hardware or manual reprogramming.

How much funding has Mowito raised? Mowito raised a $3 million pre-seed round announced on July 7, 2026, led by Version One Ventures, with participation from All In Capital, Unisol, iSeed, and angel investors including Soumith Chintala.

Who are Mowito’s co-founders? Mowito was co-founded by Puru Rastogi (CEO), Safar V (CTO, an IIT Madras graduate), and Adityanag Nagesh (Chief Business Officer, a three-time founder).

Which companies use Mowito’s technology? Mowito has deployed its robotics software in production at Denso and a leading global EMS (electronics manufacturing services) provider

What is “physical AI”? Physical AI refers to artificial intelligence systems designed to perceive and act in the real, physical world, such as robots that use vision and sensor data to manipulate objects, as opposed to AI systems that only process text, code, or digital data.

Where is Mowito based? Mowito operates from Bengaluru, India, and Detroit, Michigan, reflecting its dual focus on engineering talent in India and proximity to U.S. automotive manufacturing customers.

Who advises Mowito? Mowito’s advisors include Soumith Chintala, CTO of Thinking Machines Lab and co-creator of PyTorch, and Lerrel Pinto, a robotics researcher recognised on MIT Technology Review’s TR35 list.

Conclusion

Puru Rastogi Mowito represents a specific, disciplined bet inside a robotics funding boom full of ambitious but unproven ideas, not unlike the enterprise-AI wager behind Runlayer’s own funding run: that the fastest way to change manufacturing isn’t building new robots, but making the ones already on the factory floor smarter. With a $3 million pre-seed round behind it, named production customers including Denso and a top-tier EMS provider, and a technical advisory board few pre-seed startups can match, Mowito has more real-world validation than most companies at this stage. Whether it can scale that early traction into a defensible, category-defining business in physical AI is the story worth watching next.