The Maker of a Popular Kids Robot Just Opened a Physical AI Stack With a Built-In Safety Referee

Inner Sky Labs, the Mumbai company best known for Miko, a companion robot for children sold in more than 140 countries, has opened the AI stack behind it as a general-purpose physical AI platform for robots, drones and other machines. The announcement came on October 8, alongside news that the company is closing a 3-billion-rupee funding round — about $30.9 million — with existing backers YourNest Venture Capital and Keshav R. Murugesh joined by undisclosed global investors.

The platform splits into perception, action and safety. Drishti models handle vision, Shruti processes audio and Sparsh covers touch. On the action side, Kriya, Prana and Karma plan and execute movements — reaching, gripping, walking, flying — while attempting to predict what happens next, such as whether a box might slip out of a gripper or whether a person is about to step into a robot's path.

The most closely argued piece is Kavach, a separate safety layer that checks a machine's planned movement against operating rules before allowing it to run. For a robotic arm, those checks cover whether people are nearby, whether the speed is appropriate and whether the arm stays inside its permitted workspace. Customers and regulators can define their own rules, which matters for factories that have to satisfy local safety law rather than a vendor's defaults.

Deployment is deliberately flexible: cloud API, on-premises or on-device, with support for offline and air-gapped environments. The perception models, action models and Kavach layer can be licensed separately. Access is by invitation to qualified partners for now, and the company says the new money will go toward real-world data collection, compute and model training.

The bet is that a decade of consumer robotics translates into industrial credibility. Miko has shipped more than 500,000 units, and the company points to an independent evaluation by the cybersecurity practice of BDO in which the Miko safety system caught every unsafe prompt across seven public AI safety benchmarks and ranked first against four widely used commercial models. That result concerns conversation safety rather than physical motion — a different risk profile, as the company itself concedes.

"Every robot company will eventually have to answer one question: why should anyone trust this machine near a person," said Sneh Vaswani, co-founder and chief executive of Inner Sky Labs. "For ten years, our answer was Miko — a robot trusted enough to be left with a child. Today we are answering that same question for every kind of machine."

The pitch lands Inner Sky Labs in a crowded lane: physical AI foundation models are being built by NVIDIA, Google DeepMind, AWS and a long line of startups, most with far larger budgets. What Inner Sky has that most do not is a shipped consumer product whose safety layer had to survive contact with children — and, if the round closes as described, roughly $100 million raised to prove that experience transfers to factory floors and drone fleets. External deployment evidence, not benchmark claims, will decide whether it does.