You push, it steers: blind walkers take a wheeled AI guide onto their own streets

Problem

When Casey Mathews switches on his Glide, it finishes booting and says two words: “Let’s glide!” Mathews, a longtime white cane user, wrote that “those two words have been a long time coming for me.” He paid early to be one of Glidance’s first customers, a group the company calls Founding Gliders.

Blind people who travel alone mostly rely on a long white cane or a guide dog, and guide dogs are scarce. Citing the guide dog school Guiding Eyes, Popular Science reported in 2024 that only 2% of blind and partially sighted people in the US had one in 2017. Breeding and training a dog can cost more than $50,000, and only about half of puppies pass.

Amos Miller, Glidance’s founder and CEO, was diagnosed at age 5 with retinitis pigmentosa, a genetic eye condition, and lost most of his sight in his twenties. “I’m not trying to compete with the dog. I’m trying to provide a solution for people who don’t use them,” he told Popular Science in 2024.

AI solution

Glide has two all-terrain wheels at the bottom of an adjustable, telescoping handle. Mathews sums it up in five words: “You push, and Glide steers.” The walker sets the pace. Glide steers and brakes, and pushes back against the hand when the way ahead is not safe.

Two stereo-depth cameras in the handle work like a pair of eyes, judging distance. AI software on the device picks out obstacles, curbs, stairs, drop-offs, overhead hazards and moving people. At a crosswalk, Glide stops before the curb, says “approaching road” and holds a straight line while the user crosses. It does not read traffic lights: the person decides when to go.

A voice assistant, the Glide Agent, can describe the scene and lead the walker to a door, elevator or counter it can see. Obstacle avoidance works offline; maps and the agent’s cloud functions use Wi-Fi or a built-in mobile data connection. Glidance says its system needs no detailed maps, and that every device’s walks feed a shared model that improves through over-the-air updates. It has published no benchmarks or research papers.

At first the user chooses where to go and Glide handles the path; turn-by-turn directions with maps are promised in a later update. The company’s technical guide says Glide weighs about 9 pounds (4 kg) and runs for six hours or more. It cannot bear weight, so it is not for people who use a walker. It does not work in complete darkness, and it is not yet optimized for snow or ice.

Impact

Glide is a pilot. On September 16, 2026, Miller wrote: “In the past few days, for the first time, Gliders other than me have been walking solo with Glide as their primary mobility aid.” They were walking their own routes, including getting on and off a bus.

Staff work with each early user in person, starting with two people and doubling as devices and results allow. Then come devices for everyone who preordered, and later open sales.

Mathews’ review, posted the same week, is the most detailed user account we found. He states its limits himself: a handful of walks, all around his own house and neighborhood.

Outdoors, early on, Glide ran him into a mailbox and caught him on a tree. On a second unit with the same hardware and newer software, those clearance problems “improved markedly.” After that, he wrote, the only other ways he had ever walked that quickly were holding a sighted guide’s arm or a dog’s harness.

Crossings asked more of him. Glide holds whatever line the user starts on, so a small error at the near curb grows to several feet by the far side. Because Glide reacts to everything it sees, walking could feel “a little herky jerky.” He still reaches for his cane to find edges, to feel his way when he has to lift Glide around something, and to work out where he is after a crossing goes wrong. “A cane investigates and recovers better, while Glide handles continuous travel better,” he wrote.

His verdict is still yes, especially for people who like getting out but find walking with a cane a chore. His account is not independent: he paid early, and Glidance reposted it. No outside trial or safety study has been published.

The preorder price is $1,499 plus a required $30 a month. The final price is not set, and US insurance does not cover Glide for now. Mathews calls it “a real investment with ongoing subscription costs”.

People

Miller spent more than 15 years as a product executive at Microsoft Research, where he founded the Soundscape navigation app (now VoiceVista). He also chaired the board of Guide Dogs for the Blind in the UK. Glidance raised a $1.5 million early investment round in July 2024.

The team of about 25 people across the US, Canada, Slovenia and Mexico includes cofounder Kyle Buzzard and Ellie Carlson, an orientation and mobility specialist (the profession that teaches blind people to travel safely on their own). Carlson walked Mathews’ streets with him, crossings included. The company says more than 8,000 blind and low-vision people helped shape the design. Mathews points to touches like a Braille letter P on the power button.

Getting here took longer than planned. A test version was expected in 2024, then the launch slipped to spring 2026, and the first devices reached users in September 2026. Doubts remain: a spokesperson for the National Federation of the Blind told Popular Science in 2024 that many past inventions had tried to take the place of canes and guide dogs and failed.

Why it matters

Very few blind people in the US have a guide dog, so for most the choice comes down to a cane. Glide, built by a blind founder, could give people a third way to get out on their own. So far the evidence is one company’s pilot and one user’s early walks, mailbox and tree included, with newer software that made those misses rarer. The next steps are more users, an outside evaluation and a price more people can afford.

What we could not confirm (6)

• How many people have received a Glide in the pilot. The company’s plan says it started with two and will double from there, but it has published no count.
• Any independent test of Glide’s safety or effectiveness. We found no outside trial, safety data or peer-reviewed study.
• The final retail price, the subscription cost after launch, and when the production rollout and open sales will begin.
• The company’s total funding after its $1.5 million pre-seed round in 2024.
• The figure of more than 8,000 blind and low-vision co-designers comes from Glidance itself.
• Casey Mathews’ review covers only a few walks on familiar routes. He paid for an early device, and Glidance reposted his review, so it is not an independent evaluation.