The Assistive Tech Ceiling: Envision Reached a Million People and Still Couldn't Survive

On October 6, the Dutch accessibility company Envision told its community the end may be near. Unless it finds a partner to carry the company forward, active development of its apps and smart glasses will stop on December 31, 2026. A final update will move as many features as possible from the cloud onto the device itself, the app will remain free, online features will wind down with advance notice, and any active subscriptions will be cancelled before that happens so nobody keeps paying for something that no longer works.

The announcement, signed by co-founders Karthik Mahadevan and Karthik Kannan, reads less like a standard wind-down notice and more like a postmortem the company chose to write itself. Envision was an early pioneer of AI-powered assistive technology: its smartphone app reached nearly a million people worldwide, and in 2020 it launched the first AI glasses built for people who are blind or have low vision. Instead of burying the reason for the shutdown in vague language about headwinds, the founders published the numbers behind the decision. Those numbers deserve a wider audience, because they describe a structural problem in the assistive technology market that goes well beyond one company.

What Envision built

Envision started in 2017, when the two Karthiks were, in their own telling, naive enough to believe perseverance could fix the economics of accessibility. They shipped a free smartphone app that could read text, recognize faces, and identify objects aloud, with a paid pro tier for advanced features. The app eventually touched nearly a million people. The glasses that followed were the first of their kind: a camera on the face, AI in the cloud, and a voice in the ear describing the world, from restaurant menus to road signs. Along the way, the company partnered with Google, Meta, and Microsoft, collected industry awards, and even claims to have outsold a unicorn-valued competitor on a fraction of the budget. On the surface, Envision looked like a success story.

The business model experiment

The founders say the business looked healthy from the outside, with stable revenue. Underneath, the model was unraveling in two places. The free app had a pro tier with around 15,000 paying subscribers at its peak, yet each new subscriber cost more to acquire than the last. So the company made the app free and let hardware carry the business, selling the glasses while the app served as the engine. That worked for years, but glasses are a one-time sale with no recurring revenue, and consumer smart glasses were arriving at scale, on track to become as cheap and common as smartphones. (The industry's hardware-first answer looks very different: Cearvol's touchscreen charging case shows what dedicated-device design can do, even as the economics stay punishing.)

Envision's answer was Ally, a freemium software app built to run on any general-purpose smart glasses on the market: stop fighting the hardware curve and sell the software that makes commodity glasses useful. The company also tried selling to businesses instead of consumers, winning paid pilots with companies like Unilever and several museums, but found that enterprise buyers wanted scale numbers and ROI in ways that clashed with impact, and that building for businesses was not what the founders wanted to do. A push into the older-adult market went reasonably well but meant building a different product for a different audience, which they could not do well alongside the original one.

Envision's founders put the ceiling in writing: about a million reachable users worldwide, with a paying slice much smaller than that. The addressable market on paper is 300 million blind and low-vision people. The gap between those two numbers is the entire story.

The ceiling

This is where the founders' honesty gets uncomfortable. They noticed a pattern they had seen before: the first 100,000 users arrive easily, every user after that is harder, and somewhere around a million, growth slows to a crawl. When they compared notes with others in the field, they found the same ceiling everywhere. On paper, the market is 300 million blind and low-vision people. But for the majority of them, the problems that come first are ones technology cannot solve yet: poverty, lack of basic access, unemployment. That is why even products that are cheap or free struggle to get past the million-user mark, and why the share of users who can pay is a much smaller slice still.

The implication cuts both ways. It is not that demand for assistive technology is small; it is that willingness to pay is structurally capped by the economics of the population it serves. A company can reach a million people, make a product those people love, and still not generate enough recurring revenue to sustain a team of engineers. That is a hard place to build a venture-backed business, which is why the founders concluded Envision needs a different kind of owner: someone who can take it in a new direction, whether that means a new audience or a different business model.

What comes next, and the bigger lesson

The founders are explicit about one thing: this is not a eulogy for the field. Their parting argument is that the most important accessibility story of the last few years is not any single company, Envision included. It is that AI is making software accessible by default, in ways they could not have imagined when they started in 2017: on the web, on phones, and increasingly on the same commodity smart glasses they were pivoting Ally to support. The future they point to is one where the tools everyone uses also work for people with disabilities, which, they argue, would do more for blind and low-vision people than any single specialized product ever could.

That argument lands harder next to the community's reaction. In a widely shared essay, podcaster Anthony Corona, a longtime Envision glasses user, used the announcement to make a point about hype: the field needs to get better at separating what a product can do today from what its developers hope it might do tomorrow, especially when people are asked to invest their money and independence. Mainstream technology has always required compromises from disabled users, he noted, but specialized technology has its own: price, scale, support, manufacturing, a smaller customer base, and ultimately sustainability. Neither approach is inherently wrong. The evaluation has to be grounded in what actually ships.

Envision has not closed yet; the December 31 deadline is contingent on partner talks the founders say are ongoing. But they have already shipped something more durable than a final update. They published the economics of their own failure, in public, with numbers attached, and told the next generation of accessibility builders to treat the ceiling as a challenge, not a warning. That is an unusual kind of transparency in a technology industry that usually prefers its postmortems sanitized. It is also, in perspective, the most useful thing Envision could have shipped this year.