For most of Luminvera’s short life, the pitch came with an AR wearable. The company, incorporated in March 2026 and run out of Silicon Valley, started as a bet that the right device could pull an industrial engineer out of what its founder, Lu Yang, likes to call the “2D Stone Age.” A flat monitor renders a machine as lines on a plane. The machine is not a plane. The wearable was supposed to close that gap, letting an engineer see the thing in three dimensions, hands free, on the factory floor. Yang would demo it like a Tony Stark from Iron Man: here is the flat drawing, here is the same part standing in the air in front of you.
Recently, she dropped it.
“Following the pivot, we will focus exclusively on software for B2B customers in the robotics industry,” Yang says. The sentence is flat. The decision behind it runs back through most of a decade, to a problem she has been describing the whole time.
Yang built her career in the automotive industry, at Bosch, the German industrial and electronics conglomerate, and Mercedes-Benz. Trained as an IT project manager, she was responsible at Bosch for digital transformation across engineering and quality management. The job, stripped down, was watching expensive engineers do work that did not need their expertise.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
That is the image she returns to. “Imagine you are building a giant, super-cool 3D LEGO set, but the instructions are written on 30,000 flat, boring pieces of paper with tiny text,” she says. “It would be a total nightmare. You would spend more time getting frustrated than actually building.” A customer specification lands as several thousand pages of prose. Someone senior translates it, by hand, into a design. Plants and engineering centers reconcile their work through physical printouts. Compliance reports get compiled, by hand, from three-dimensional systems into flat text and 2D images.
Her complaint was never really about the tools. “Technology is usually made in a way that forces humans to change how they work, instead of the technology changing to help the human,” she says. That is the line that survived the pivot, and probably caused it.
The first Luminvera answered with hardware and an AI layer, a full-stack attempt to carry the spatial context a desktop screen throws away. The new Luminvera keeps the thesis and drops the wearable. What it sells now is software, what Yang describes as “a magical, video-game-like workspace where flat instructions instantly turn into real 3D objects you can touch and move with your hands.” An AI layer reads the thousand-page specification and turns it into structured constraints an engineer can design against. A spatial layer renders the result as something closer to an object than a drawing, so the engineer reasons in three dimensions instead of rebuilding them in her head from a flat screen.
Robots, specifically. The narrowing from “industrial engineering” to “the robotics industry” is the part of the pivot to take seriously. Robotics is where the new manufacturing money is going, and where the distance between a CAD file and a finished machine is widest and most expensive to cross. A company selling an immersive design surface has a cleaner story to tell the firms building humanoid robots than to the whole sprawl of heavy engineering at once.
Dropping the hardware follows the same logic, and the timing is not academic. Software installs on a customer’s existing machines. An AR wearable has to be built, certified, supported, and worn, and the market for those devices turned treacherous in late 2024, when Microsoft discontinued its HoloLens 2 wearable with no successor and a 2027 end-of-support date, stranding a generation of factory-floor AR built on top of it. Betting a young company on hardware someone else might stop making is a hard way to start.
The trouble is that the software lane Yang is merging into already has traffic. PTC, the Boston industrial-software company, sells Vuforia Expert Capture, an AR tool that turns 3D CAD into guided work instructions and AI-checked inspections; one industry analyst has ranked PTC the top AR vendor four years running. Scope AR, a San Francisco enterprise-AR firm whose WorkLink platform overlays CAD files onto real equipment, was acquired this month by Flatirons Solutions and folded into a larger technical-content company, a reminder of how hard the standalone path has been. Augmentir leads with generative AI, its Augie assistant spinning existing manuals and videos into digital procedures. All three are more established and better funded, and already inside the factories Yang wants. None of them, though, is built around what Luminvera is built around: the engineer at the design stage rather than the technician on the floor, and robots in particular rather than manufacturing in general. Whether that is a real moat or a narrow one is the open question.
Yang made the case in public last week. On June 11 she was among the founders pitching at the Founder Institute’s Silicon Valley Spring 2026 graduation, an online showcase of the accelerator’s newest companies. The Founder Institute, headquartered in Silicon Valley and running chapters in roughly a hundred countries since 2009, is one of the highest-volume pre-seed programs anywhere; it says its founders have raised more than two billion dollars. Its best-known graduate is Udemy, the online-learning company that went public on the Nasdaq in 2021. More recent alumni include Esusu, a rent-reporting startup that crossed a billion-dollar valuation late last year.
The graduation keynote belonged to Torrey Smith, co-founder of Endiatx, another Founder Institute company and the maker of PillBot, a swallowable micro-robot that streams video from inside a patient’s stomach and is now in trials with the Mayo Clinic. It is also a robotics company, which made Yang’s three-minute slot a fair preview of the room she now wants to sell into.
She has been sharpening this argument for years, and the pieces she leads with have moved. The “2D Stone Age,” the thousand-page specs, the printouts: those stay. The AR wearable is gone, and the word “robotics” is carrying weight it was not carrying six months ago. What Yang has, that the incumbents crowding her lane do not, is years spent as the engineer on the wrong side of the desk, translating 30,000 pages by hand. She is betting that the memory is worth more than their head start.
Get the TNW newsletter
Get the most important tech news in your inbox each week.
Opinions expressed by TNW authors are their own