1X CEO Details NEO's 'Two Modes' and Defends Teleoperation as 'More Secure' than a Cleaner

Facing intense scrutiny over its human-in-the-loop strategy , 1X CEO Bernt Børnich appeared on the New York Times podcast Hard Fork to defend the company's approach, arguing its teleoperation system is a necessary learning tool and can be "more secure" than hiring a human cleaning service.

In the interview, Børnich offered the clearest distinction to date between NEO's autonomous capabilities and its human-assisted "Expert Mode", which has been a flashpoint for critics.

"Everything I talk about here now happens through a mixture of autonomy and teleoperation," Børnich explained, describing the chores his own NEO performs, such as tidying, vacuuming, and imperfectly folding laundry. He compared this "human-in-the-loop" system to both Waymo vehicles needing occasional remote assistance and users refining prompts for ChatGPT. "It is going to be a while until there is absolutely no human in the loop ever," he stated.

NEO's "Two Modes"

Børnich clarified that NEO effectively has two distinct operating modes.

• 1. The "Mixture" (Chores Mode): This is the primary mode for completing scheduled tasks like cleaning. It blends autonomy with teleoperation, allowing a remote operator to assist when the robot gets stuck. This is the system that provides "expert demonstrations" to train the AI.
• 2. The "Fully Autonomous Mode": This mode does not allow for human intervention. Børnich claimed that in this mode, NEO can already function as a conversational AI companion , open doors with "90 something percent" success , and fetch objects with "maybe... 80% successful" rates.

This framing attempts to counter critiques, like those from Marques Brownlee, that 1X is "selling the dream" of autonomy while only delivering teleoperation. Børnich’s argument is that the teleoperation is the engine for the autonomous mode, not a replacement for it.

The Data-Gathering Gamble

The CEO was transparent that NEO is "a data collection play", designed to solve the industry's "chicken-and-egg problem" by gathering real-world data that doesn't exist online.

Børnich argued that the internet is a "bad representation of human life" and that real-world deployment is the only way to "live and learn among us". To achieve this, he revealed 1X plans to ship "more than 10,000 units next year", a fleet he claimed is larger than Waymo's and will generate a data volume approaching that of YouTube's daily uploads.

He also frankly admitted the financial stakes of this "data flywheel" model. He argued that paying for teleoperators is currently cheaper than the "alternative cost" of building massive, warehouse-sized "mock up kitchens" to gather data, as some competitors are doing.

However, he conceded that this model is not sustainable forever. "There's no doubt that you're running towards a cliff," Børnich said, explaining that 1X must "get more and more autonomy working as you scale" or the company "will at some point need to like stop scaling until the autonomy works".

A "More Secure" Cleaning Service?

Addressing the significant privacy concerns of having a data-collecting robot in the home, Børnich compared NEO's service directly to a traditional cleaning service. He argued 1X can provide a "better, more secure service" through several layers of protection:

• Operator Vetting: All remote operators are vetted.
• Active Oversight: A manager monitors operators in real-time, with one manager overseeing eight teleoperators.
• Audit Trail: Video logs are kept of all operations to trace any incidents.
• Anonymization: Operators use a "Starcraft-like" interface to queue tasks. They cannot see people (who are "blurred out") and do not know which home they are operating in.
• Task-Specific VR: Only when a task fails is it delegated to an operator in VR, who sees only the specific task (e.g., a jacket to be folded) before disconnecting.

Børnich also stated that NEO only collects data (video, audio, and touch/force) when it is "active and... doing something" , and that data is stored locally first, allowing users to delete it before it is sent to the secure cloud.

A "Clunky" Demo

Following the interview, the Hard Fork hosts conducted their own demo with NEO in their office kitchen. The robot, which was 100% teleoperated by an employee in another room, successfully gave a hug, retrieved a drink from a fridge, and filled a cup of water.

However, the demo also highlighted the hardware's limitations, echoing clumsy moments seen in 1X's "Culinary Bootcamp" video. NEO dropped a pair of tongs it was asked to put away and, most notably, "could not" pick up again from the floor. An attempt to squat caused the robot to become "off balance" and "almost looked like it was going to fall backward," requiring a human to stabilize it.

A 1X representative blamed the failures on "Wi-Fi issues" at the office, stating the robot was "not properly calibrated".

The hosts concluded that NEO is currently in a "real buyer-beware situation", comparing the robot not to a butler but to "an intern that needs a lot of your attention and needs to be trained up".

The 2027 Autonomy Target

Despite the demo's mixed results, Børnich laid out an aggressive timeline for the future. He projected that NEO could achieve full autonomy on par with a human cleaner by 2027 in a "bullish" scenario, or 2028-2029 in a "bear case".

He also predicted that humanoid robots could achieve 10% penetration in American homes by "2030, maybe," suggesting the primary bottleneck will be "a manufacturing problem," not a lack of demand.

Watch the interview below: