Three things worth knowing
- $3M, and it's a pre-seed round, not a seed round. Version One Ventures led, with All In Capital, Unisol and iSeed participating, alongside angels including PyTorch creator Soumith Chintala.
- The model learns from demonstration, not code. Mowito says its foundation model can pick up a new robot-arm task from as little as one human demonstration, and runs on standard industrial arms without hardware modification.
- Three job categories, per Mowito's own product pages: precision assembly and pick-and-place, vision-and-tactile inspection, and machine tending and logistics.
- It's already in production at a Fortune 500 automotive company and one of the world's largest electronics contract manufacturers, per the funding announcement.
- Same week, different vertical: Chennai-founded Blurgs AI raised $2.2M for maritime and defence intelligence — two very different embodied-AI bets from Indian founders within a single week.
- The backdrop is a record year. Global robotics-startup funding has already hit $18.8B in 2026, according to Crunchbase data, ahead of the whole of 2025 with six months still to run.
Inside the round: who backed Mowito, and why
Mowito, co-founded in 2024 by Puru Rastogi (CEO), Safar V (CTO) and Adityanag Nagesh (CBO), announced on 7 July 2026 that it had closed $3M in pre-seed funding led by Version One Ventures. All In Capital, Unisol and iSeed joined the round, alongside a notable angel bench: Soumith Chintala, creator of PyTorch — the framework underpinning most of the world's production AI systems — plus Adarsh Kulkarni, Ashish Kulkarni and Vaibhav Domkundwar. The company operates out of Bengaluru and Detroit, and plans to use the fresh capital to expand its US presence, build out its Detroit engineering and go-to-market team, and deepen deployments with automotive and electronics manufacturers.
In Version One Ventures' own words, announcing the deal: "Instead of rigid programming, Mowito's model learns a task from as little as a single demonstration." That is the pitch in one line — industrial robot arms have historically needed engineers to hand-code waypoints and rewrite logic for every new part, fixture or product variant. Mowito's bet is that a foundation model trained across many factory tasks can absorb a new one the way a person would: shown once, then corrected.
| Detail | What's confirmed |
|---|---|
| Round size | $3M, pre-seed |
| Lead investor | Version One Ventures |
| Other investors | All In Capital, Unisol, iSeed |
| Notable angel | Soumith Chintala (creator of PyTorch) |
| Founders | Puru Rastogi (CEO), Safar V (CTO), Adityanag Nagesh (CBO) |
| Founded | 2024 |
| Headquarters | Bengaluru and Detroit |
| Announced | 7 July 2026 |
| Use of funds | US expansion, Detroit team, automotive/electronics go-to-market |
When a pre-seed round is led by a fund with a track record in developer tooling and physical AI, and joined by a named technical angel such as a widely-used framework's creator, treat that as a stronger signal of product substance than the round size alone. At this stage, the cap table is a proxy for who actually did the technical diligence — worth checking before you read too much into a headline dollar figure.
What Mowito's foundation model actually does
Strip away the "physical AI" label and the product is concrete: a vision- and tactile/force-guided system that drives existing robot arms through tasks that used to require fixed jigs, rigid programming and a specialist integrator for every product change. Mowito's own site groups the work into three categories, each aimed at a specific factory-floor pain point.
| Task category | What it replaces | How Mowito describes it |
|---|---|---|
| Precision assembly & pick-and-place | Fixed fixtures, hard-coded waypoints, per-product reprogramming | High-precision part assembly and fluid dispensing without fixed supports, adaptable to moving conveyors |
| Inspection | Multiple fixed cameras, manual QA sampling | A single arm-mounted camera doing multi-angle, high-speed defect and dimensional-error detection |
| Machine tending & logistics | Dedicated loaders per machine, manual handling between stations | Mobile robotic arms managing loading and unloading across multiple machines |
The mechanism is what makes this a "foundation model" story rather than a conventional robotics-integrator story: instead of a bespoke program per part number, the model is trained broadly enough that a new task transfers from a small number of demonstrations, and the deployment loop — configure the setup, train on site-specific data, deploy the update — runs largely through software rather than a re-engineering project. Mowito says its robots are already running production lines for a Fortune 500 automotive company and one of the world's largest electronics contract manufacturers, and its own site lists automotive and electronics names including DENSO among its customer and partner logos.
If you're evaluating a demonstration-taught robotics vendor for a production line, ask for the failure mode, not just the headline accuracy number. A system that hits near-100% reliability on a stable part but degrades sharply on a minor tolerance change tells you far more about real-world readiness than a single benchmark figure.
Mowito is a pre-seed company with two publicly named reference deployments. That's a genuine signal at this stage, but it is not the same as an independently audited reliability track record across hundreds of factories. Treat vendor-reported accuracy and cost-reduction figures as directional until third-party deployments accumulate.
Same week, different vertical: Blurgs AI's $2.2M for maritime and defence intelligence
Mowito's round did not land in isolation. Within a day of its announcement, Blurgs AI — founded by IIT Madras alumni Roshan Raj Mohanty and Dr Avinash Kori, who holds a PhD in AI from Imperial College London — raised $2.2M led by Pravega Ventures and Shastra VC, with angels including PlaySimple Games co-founder Suraj Nalin and Fyle co-founder Yashwant Madhusudhan. Blurgs AI is a different kind of physical-AI company: rather than manipulating objects with a robot arm, it builds sensing and situational-awareness platforms for ports, fleets, shipyards and coastal defence, serving named clients including the Indian Navy, Indian Coast Guard, Bharat Electronics, DRDO laboratories, Mumbai Port Authority and Dubai Maritime City.
Put side by side, the two rounds are a useful snapshot of how broad the "physical AI" label has become — one startup is teaching a robot arm to pick up a part, the other is teaching a sensor network to track a vessel. Both depend on the same underlying shift: AI systems that reason about and act on the physical world, not just text and tokens.
| Lens | Mowito | Blurgs AI |
|---|---|---|
| Round size | $3M pre-seed | $2.2M |
| Lead investor(s) | Version One Ventures | Pravega Ventures, Shastra VC |
| Founders' background | Carnegie Mellon, IIT Madras, serial founder | IIT Madras, Imperial College London (PhD) |
| Physical-AI vertical | Robot-arm manipulation (assembly, inspection, tending) | Maritime & defence sensing, situational awareness |
| Named customers | Fortune 500 automaker; large electronics manufacturer | Indian Navy, Coast Guard, Mumbai Port Authority |
| Announced | 7 July 2026 | 7-8 July 2026 |
Physical AI's record year — and why India's manufacturing base is a genuine edge
Both rounds arrive inside a broader funding surge. Robotics startups have already raised $18.8B globally in 2026, according to Crunchbase data reported by Crunchbase News — ahead of the $15B raised across all of 2025 and the previous 2021 peak of $14.1B, with half the year still to go. The headline rounds are enormous by comparison to Mowito's: autonomous-maritime-defence startup Saronic raised a $1.75B Series D, Germany's Neura Robotics pulled in $1.4B for humanoids, and Skild AI raised $1.4B to build an "omni-bodied" brain that can operate any robot. Investors are increasingly treating physical AI as a category on the same scale as foundation-model labs, not a niche hardware bet.
For builders in India, that is the real story underneath the $3M number. As general-purpose chat models commoditise and the cost of a capable LLM API call keeps falling, the differentiated ground shifts toward domains where physical deployment, not just model quality, is the hard part. India's manufacturing base — automotive component makers, electronics contract manufacturers, and a growing base of precision-engineering exporters — gives Indian founders and engineers a distinctive vertical to build in, rather than competing purely on general intelligence with labs an order of magnitude better funded. Genesis AI's work moving physical AI out of the lab and open robot-reasoning models such as MolmoAct2 point the same direction: the frontier of usable physical AI is being built by teams who combine model research with unglamorous, on-site deployment work.
That combination — model plus deployment — is also why this fits the wider pattern in India's AI capital surge, where the deals attracting serious cheques increasingly sit in applied verticals rather than general-purpose model training, and why the broader record VC year for AI startups has room for a $3M pre-seed round in industrial robotics alongside billion-dollar foundation-model raises.
Every article here is written for Verified Builders. Want your name on the next one?
AI Tech Connect lists AI engineers, founders and researchers across India and the UK — and the people hiring browse it to find them. Adding your profile is free.
Become a Verified Builder →The UK angle: an automation gap British builders can move into
This is an India-first story, but it isn't an India-only one. Britain has its own version of the physical-AI opportunity, and arguably a starker gap to close. The UK has the lowest robot density of any G7 economy, according to figures cited by the UK government's own robotics and automation guidance — a data point that has circulated for several years but still holds. In response, the government has committed £4.3bn to advanced manufacturing over five years, including £2.8bn earmarked specifically for R&D in automation, robotics and smart factories.
That is real money chasing the same underlying problem Mowito is solving in Bengaluru: too few skilled integrators, too much bespoke engineering per production line, too little AI actually touching the factory floor. British manufacturing — automotive, aerospace, precision components — is a comparable vertical to India's, and the £8.2bn UK AI start-ups raised in H1 2026 shows capital is available for builders who can pair AI research with physical deployment skills, not just another SaaS wrapper. London Tech Week's £6bn of announced AI investment this year leaned heavily toward applied and infrastructure plays — the same shift Mowito and Blurgs AI represent in India.
| Lens | India | UK |
|---|---|---|
| This week's signal | Mowito ($3M) and Blurgs AI ($2.2M) pre-seed rounds | £8.2bn raised by UK AI start-ups in H1 2026 |
| Structural gap | Manufacturing scale without enough AI-native integration | Lowest robot density in the G7 |
| Public backing | IndiaAI Mission compute subsidies | £4.3bn Advanced Manufacturing Plan (£2.8bn for automation R&D) |
| Builder opportunity | Vertical-specific physical AI for automotive/electronics | Retrofitting AI onto an ageing manufacturing base |
What builders should actually do with this
A $3M pre-seed round is easy to skim past as a small line in a funding roundup. The more useful question is what it signals about where the next few years of physical-AI hiring and building actually happen.
- Physical AI rewards deployment skills, not just model skills. Vision-language-action models, sim-to-real transfer, and basic ROS/robot-integration knowledge are becoming as valuable as prompt or fine-tuning expertise — arguably more so, because far fewer people have them.
- Show shipped work, not a job title. Companies at Mowito and Blurgs AI's stage hire on evidence of things that actually ran on real hardware or real data, not a CV line claiming "robotics experience."
- Don't treat this as an India-only category. The UK's manufacturing base and its £2.8bn automation R&D commitment mean the same skill set — AI plus physical deployment — is investable and hireable on both sides of this site's audience, not just in Bengaluru or Detroit.
None of that requires waiting for the next funding headline. A public, verifiable Builder profile is the fastest way to be visible to a founder scanning for exactly this combination of skills before a role is even posted — and Founding Builder spots carry a badge that's visible to exactly the people looking for it, while they remain open.
Primary sources: Version One Ventures' investment announcement, Mowito's own product pages, and The Week's reporting on Blurgs AI.