OpenAI-backed 1X deploys humanoid guards ahead of Tesla

While Elon Musk has been hyping Tesla’s Optimus robot, an OpenAI-backed startup just quietly beat him to a real-world commercial deployment.

1X, a robotics firm backed by the creators of ChatGPT, has successfully deployed its wheeled humanoid robot, EVE, into professional work environments. According to reports in late May, EVE has been on the clock since April as a security guard at two industrial facilities across the US and Europe.

Security guards with human hands

Egg-shaped security drones like Knightscope’s K5 and two-wheeled patrol bots like Ascento have been rolling around for a few years. But EVE looks and works fundamentally differently.

EVE features a head, a digital face, and two fully articulated manipulators. Because it has functional arms, it doesn’t just record video; it can actually turn door handles, open windows, and pick up dropped objects on its route.

1X founder and CEO Bernt Bornich calls previous humanoids mere gimmicks that never survived outside the lab. Security work gave 1X a practical starting point with a narrow scope of responsibilities.

Human security guards monitor the fleet. If a robot gets confused or encounters an unexpected obstacle, human operators can virtually step in and pilot EVE remotely. Every patrol and manual intervention feeds real-world training data back into the system so the robots can perform more tasks autonomously over time.

From warehouses to elder care

Patrolling industrial sites is just the baseline. EVE has already run pilot tests in retail stores, logistics hubs, and healthcare facilities, where people naturally cleared paths for the rolling machines.

1X is now preparing to deploy EVE to hospices and assisted living facilities as caregivers for the elderly. The company has tested EVE on nursing duties and even bartending.

Meanwhile, 1X is already developing a fully bipedal humanoid dubbed NEO, which trades the wheeled base for human-like walking legs. Bornich believes that gathering massive amounts of real-world interaction data could help solve acute labor shortages in sectors like security—which employs 1.1 million guards in the US alone—within the next five to ten years.

Why it matters

We think 1X made a brilliant tactical compromise that puts most humanoid robotics labs to shame.

Most teams fall into the trap of trying to build a perfect bipedal robot from day one, spending years chasing balance algorithms while their machines remain trapped in demo videos. 1X basically said: let’s put human arms on wheels. That hybrid design sacrifices a bit of sci-fi aesthetics for immediate commercial utility, allowing EVE to enter paying industrial environments right now without tripping over its own feet.

Crucially, 1X turned security patrolling into a data-collection engine. By having human guards remotely pilot the machine whenever it gets stuck, 1X keeps clients happy while gathering the holy grail of embodied AI: long hours of messy, real-world manipulation data.

That said, the approach hits a wall once 1X tries to enter homes and hospices. A wheeled base works fine on flat warehouse floors, but carpets, steps, and cramped rooms will quickly break the illusion. Whether 1X can take the intelligence it builds on wheels and smoothly transfer it to its upcoming bipedal NEO is the real multi-billion-dollar test.