OpenTraining Joins Meta Wearables Studio UAE to Build Smart Glasses Technology That Protects Workers

The platform turns skilled work into training data for humanoid robots and pixelates every face on the device, before a single frame is stored.

ABU DHABI, UNITED ARAB EMIRATES, October 30, 2026 /EINPresswire.com/ -- OpenTraining, a training-data platform for physical AI, has been selected into Meta Wearables Studio UAE, the national programme developing AI wearable applications with security and privacy designed in from the beginning.

The programme, unveiled at GISEC Global 2026, is delivered by Startupbootcamp with Meta and the UAE Cyber Security Council, and supports the UAE National Strategy for AI 2031. It gives UAE-based developers, entrepreneurs and startups access to Meta’s wearable technology and engineering expertise, including Ray-Ban Meta and Oakley Meta smart glasses.

OpenTraining uses those glasses for a specific purpose: recording how skilled industrial work is actually performed, so that humanoid robots can eventually be trained to take on the most dangerous of it.

“Smart glasses have spent two years being discussed as a consumer novelty and a privacy worry,” said George Bancs, Founder and CEO of OpenTraining. “They are neither. In the hands of a skilled worker they are the most useful industrial recording instrument ever made, because they see exactly what the expert sees. The question was never whether to allow them. It was what we choose to build with them.”

On OpenTraining’s platform, face detection and pixelation run on the device in real time. The system writes the redacted image rather than the camera feed, so identifiable frames are never stored, never transmitted and never reviewed by a person. Licence plates and documents run through the same pipeline.

“Privacy by design is not a feature we added, it is the condition that lets this technology onto a worksite at all,” Bancs said. “No operator will let a camera near their people or their assets if the footage can travel somewhere they cannot control. Build the protection into the device and the question stops being whether this is safe and starts being what it can do.”

That approach aligns with the principle the programme was founded on. Dr Mohamed Al Kuwaiti, Head of Cyber Security for the UAE Government, has said that cybersecurity and privacy need to be built in early in development rather than added afterwards.

“Being inside this programme means the privacy and security work happens next to the people building the hardware, with a national cyber security council in the room,” Bancs said. “That is a materially better way to get a new technology right than debating it from the outside after it has already shipped.”

Industrial work still kills and injures people at a rate that has barely improved in a decade, and the most hazardous tasks, including confined space entry, work at height, live services and commissioning, are precisely the ones that require the deepest human judgment and are hardest to automate. That judgment has never been recorded in a form a machine can learn from.

Through the company’s Training Economy programme, skilled workers record demonstrations on smart glasses during ordinary shifts and are paid for every approved contribution, with a share of each downstream dataset licence routed back to the person who recorded it. The platform captures two tracks from every recording. The first is the task performed. The second is the safety judgment inside it: the hazard an experienced worker recognised and avoided, the shortcut refused, the load not taken from that angle.

“The people who hold this knowledge have been handing it down for free for as long as the trades have existed, and losing it entirely when they retire,” Bancs said. “We are putting a price on that transfer for the first time, and pointing the result at the work that hurts people. A machine doing confined space entry instead of a person is a good outcome. A machine doing it because an expert was paid properly to teach it is a better one.”

The safety output is immediate, and it has become the subject of most of the company’s inbound enquiries.

A capture recorded on a site in September, one minute and twelve seconds of a worker walking an unfinished floor, returned a site safety report flagging twelve hazards, graded one high, two medium and nine low. Each entry recorded what was observed, what would happen if it were left unaddressed, and the action required, timestamped to the second it appeared in the recording. The high-severity item was an unguarded penetration in the slab, with a recommendation to barrier or cover it before anyone used the route.

“That came out of seventy-two seconds of walking, with no safety officer present,” Bancs said. “The same recording that will one day teach a machine how the job is done tells the operator what is wrong with their site today. Most of our enquiries now begin there, and I think that is the right order. Make the technology useful to the people wearing it first, and the rest follows.”

Hazard profiles have been authored across sixteen industry verticals, beginning with building and infrastructure.

About Open Technology LLC

Open Technology LLC owns and operates OpenTraining.ai, a training-data platform for physical AI. OpenTraining is a member of Meta Wearables Studio UAE and a member of NVIDIA Inception, NVIDIA’s programme for AI startups.

Founder and CEO George Bancs is an inventor, author and strategist working on physical AI and the frameworks that govern it. He is a three-time industry award winner for workforce safety technology and the author of Synthanity.

Meta, Ray-Ban Meta and Oakley Meta are trademarks of Meta Platforms, Inc. NVIDIA and the NVIDIA logo are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries. Participation in Meta Wearables Studio UAE and membership of NVIDIA Inception do not imply endorsement by Meta or NVIDIA.

Press Contact
Agentic Press Holdings
connect@agenticpressholdings.com