Most product teams optimise for the median user. It is the rational thing to do — the median user is where the volume is, where the retention curve bends, where the revenue sits. But for the last four years I have been building for people the median never includes: blind and visually impaired users of the Smarton AI app and smart glasses at Sunbots Innovations, and the wider disability community I worked with at Voice of Specially Abled People.
That work changed how I think about product.
When we set out to refine the Smarton AI app, the first instinct was the usual one: pull the analytics, find the drop-off, patch the funnel. The analytics told us where people stopped. They did not tell us why.
The why only came out of on-ground survey work — sitting with users while they tried to read a printed letter, a medicine label, a government form. It turned out that our multi-format document reader was technically correct and practically unusable in the exact moment people needed it, because it assumed a steady hand and a quiet room.
That single observation reshaped the roadmap more than any dashboard did. It led to the two-way voice interaction model, and eventually to the smart glasses attachment — a camera worn on the frame, connected over a Type-C cable, so the user's hands stay free and the device scans what they are walking toward rather than what they have carefully positioned in front of a phone.
The glasses are, deliberately, not a moonshot. They are a camera, a cable, and an app that already worked. Every fancier version we sketched — wireless, on-board compute, its own battery — added cost, weight, and one more thing to charge.
Accessibility products live under real constraints: price sensitivity, low tolerance for setup friction, users who cannot troubleshoot visually when something breaks. Those constraints are not obstacles around the product. They are the product specification. The simplest thing that removes a real barrier beats the sophisticated thing that adds one.
The other lesson is organisational. On a small team, engineering, design and outreach each hold a different piece of the truth: engineering knows what is feasible this quarter, design knows where the interaction breaks, outreach knows what users said last week and what a CSR partner will actually fund.
A roadmap that only reflects one of those three ships confidently in the wrong direction. Most of my job as Head of Product is holding those three views in the same room until they agree on a sequence — not writing the sequence myself.
You do not have to work in assistive technology for this to matter. Every product has a last 15% — users on slow connections, on old devices, in a second language, under time pressure, in noisy environments. They are usually the users your metrics quietly write off.
Designing for them rarely costs what you think it will, and it almost always produces a product that is clearer for everyone else too. Voice navigation, high-contrast modes, hands-free interaction, plain language — none of these started as mainstream features. They all became them.