Creating the future, not just predicting it.

One of the things we love most at Envision is being at the bleeding edge of technology. Not just following trends, but shaping them. Playing with the new, the experimental, the not-yet-ready-for-prime-time, and turning it into something meaningful for our community.

So when Meta’s Reality Labs Research team gave us a call about their upgraded Aria Gen 2 glasses it was an easy yes. Aria Gen 2 is the second iteration of Project Aria’s research device, a not-for-sale research tool given out through an application process. What we saw was one of the most sensor-packed research glasses we’d ever encountered, all in a sleek, wearable form factor.

Naturally, the wheels started turning.

We decided to team up to build a bold prototype that combined Aria Gen 2’s spatial audio capabilities with Envision’s agentic AI, with one big goal: help blind and low-vision users navigate complex indoor spaces.

To ground the idea in reality, we turned to our community. We asked: where do you face the most challenges indoors? The top four that emerged — supermarkets, offices, airports, and train stations, didn’t surprise us, but gave us clarity.

We chose to start with supermarkets.

Specifically, navigating to and identifying items in the produce section, something deceptively simple, but extremely nuanced when done independently and accessibly.

So we got to work. Meta’s team built the spatial audio infrastructure, enabling sound to be placed in a 3D space using stereo speakers. Meanwhile, we designed a conversational AI experience that could do things like:

Recall a saved shopping list
Identify fruits and vegetables (“What kinds of apples are here?”)
Handle more nuanced queries (“Which apple is best for a pie?”)

Behind the scenes, this AI had to pick the right tool, camera, memory, web search, LLM, or navigation, to answer each question in context.

After weeks of building prototypes across the Pacific, it was time to bring it all together. So our team: Arman Soudi, Alican Aşan, Bhagya Goonewardhane, Karthik K., and I, flew down to Meta’s campus in Seattle

A group of four people are gathered in a modern conference room around a wooden table with laptops. One person on the right, wearing AR glasses, gestures toward a large screen on the wall that shows a mirrored camera feed of the meeting.

Understanding the possibilities and limitations

A person in a white shirt and light jeans is seen from behind adjusting a pair of smart glasses in a modern industrial-style hallway with exposed ducts and lighting. Three other people stand nearby — one looking at a phone, two engaged in conversation. A wooden kiosk-style desk holds a laptop displaying a dark interface.

Testing in closed settings

There, we integrated both systems on the actual device and tested it indoors. Meta had already scouted a supermarket in Carnation that was ideal for our field tests.

Now, we could’ve stopped there. We had a working prototype. We could’ve shot a polished video and called it a day. But that’s not how we roll.

Article content

Testing in the field

A person holds a foldable tablet displaying a terminal interface with lines of code or logs. In the background, one man stands with a laptop while another observes the fruit display, which includes melons, apples, and packaged produce.

Breaking and fixing things as we go

In a supermarket produce section, two men sit on the floor working on laptops in front of a display of fruits, including watermelons and melons. Behind them, another man adjusts a pair of smart glasses while a colleague watches nearby with a laptop on a wooden stand. Coffee cups and backpacks are scattered around, indicating a casual yet focused tech setup in an unconventional setting.

After a point, we really started to feel at home (office) there

We spent two full days at the supermarket ensuring the prototype worked as intended, no smoke, no mirrors. We also had Sharon Giovinazzo, the CEO of Lighthouse SF, a real user, test the experience end-to-end firsthand.

A woman wearing smart glasses and holding a white cane stands in front of a vegetable display. Two men stand nearby, one monitoring something on a smartphone connected by a cable supporting the smart glasses setup. The produce shelves behind them are stocked with fresh vegetables like carrots, kale, lettuce, and cucumbers.

Testing and validating with blind and low-vision users

A woman with a guide dog stands in front of a produce display in the organic section of a supermarket, facing shelves stocked with lettuce, cilantro, peppers, and other vegetables. She is accompanied by a man nearby holding an open laptop. The guide dog, a yellow Labrador, is wearing a harness and looks attentively ahead.

Sharon giving her final bits of feedback day before the shoot

Only once it actually worked, did we capture it on film.

It was an intense sprint, but one that taught us a lot, not just about where smartglasses are headed, but how we at Envision need to evolve with it. This wasn’t just a demo. It was a glimpse into a future we believe will soon be the norm.

And we’re not done. Envision and Meta continue to explore ways to bring more futuristic concepts to life.

Stay tuned for the next adventure.

A picture of me wearing the Aria Gen 2 glasses inside a supermarket holding a film slate labeled “Meta RL-004” with details including the names and dates of filming. I’m wearing a dark dotted shirt with the Envision logo and a blue jacket. In front of me are two black equipment cases marked “FRAGILE” and labeled with orange tapes.

It’s a wrap (until the next blockbuster)

Special thanks to Prince Gupta Saransh Solanki Chen Kong Archit Gupta for your cooperation on this 🙏