We rarely notice the sloped edge of a path, but the 'curb cut' didn't begin in a city planning office. It began as an act of guerrilla engineering.
In the early1970s, disability activists in Berkeley took sledgehammers to a curb and poured a makeshift concrete ramp. It wasn’t pretty, and it wasn’t sanctioned. But within a decade, that crude wedge of concrete had become the curb cut - a feature now legally required at virtually every street corner in the developed world, Dublin included.
The curb cut was designed for wheelchair users. But ask anyone pushing a buggy down around town, dragging a suitcase to Heuston, or rolling a shopping trolley to their car, they’ll tell you who else benefits.That’s the curb cut effect: a design fix made for one group of people, originally framed as a cost or an accommodation, ends up quietly reshaping the experience for everyone.
We see this cycle repeating in the digital world; it is perhaps the most important lessonwe have for navigating the shift toward AI-driven design.
Think of the tools you’ve used since you got out of bed.
Theauto-captions on your team’s video call started life as an accessibility feature for Deaf and hard-of-hearing users. Now they’re often switched on by default in every open-plan office and every video on every social feed.
Voice controlon your phone, “Hey Siri, set a timer”, was originally a screen reader and motor-accessibility feature. Now it’s how people cook with messy hands and how drivers navigate without taking their eyes off the road.
Dark mode was championed for years by users with light sensitivity, certain visual impairments, and migraine before it became a mainstream default. A product without it feels dated.
Even autocomplete and predictive text trace their lineage back to assistive communication devices for users with motor impairments (permanent, temporary, or situational1) and aphasia.
Every one of those features started as “accessibility work” and ended up in the mainstream product. None of them were rolled back once the wider market caught up.
There’s a useful frame from inclusive design research that’s worth pausing on here,because it’s the engine behind every example I’ve just listed. Disability isn’ta fixed category, it sits on a spectrum that almost everyone is on at some point.
A permanent impairment is what most people picture when they hear the word disability: a user with low vision, someone who’s been Deaf since childhood, a person bornwith limb difference, a wheelchair user. These are the users we most readily associate with accessibility features, and they’re the users our work islegally and ethically obliged to serve.
A temporary impairment is the same functional limitation, time-limited. A broken wrist in acast, post-surgery recovery, laryngitis before a big presentation, aconcussion. The user isn’t part of the “disability community” but for a fewweeks, their needs are functionally identical to someone who is.
A situational impairment is what nearly everyone experiences several times a day. Holding at oddler on one arm, trying to read a screen in bright sunshine, listening to avideo on mute when you've forgotten your headphones on the Luas. The user hasno disability at all but the environment has temporarily handed them one.
Here’s the interesting part: a single design fix usually serves all three. Permanent impairments affect roughly 16% of the global population on any given day.Temporary and situational impairments? Closer to 100%2. Designing for the 15% raises the floor for the 100%, that’s the curb cut effect with a different label.
Defining 'the accessibility user' as a narrow outlier is an old-school mindset that fails to account for the fluid way we all interact with technology today. Accessibilityusers aren't a segment of your audience. They're your audience at moments your analytics dashboards don't capture.
Here’s why I’m optimistic about the next five years.
The work happening right now to make AI products usable for disabled users: structuredoutputs that screen readers can parse, predictable response shapes, semanticlandmarks in generated UI, transparent reasoning that can be audited by userswith cognitive disabilities. These will define how every user interacts with AIby the end of the decade.
Think about what blind users have been asking for from chatbots: don’t bury the answer at the end of a 400-word preamble. Be predictable about where the conclusionlives. Use real headings, not bold text pretending to be headings. Tell mewhat you’re about to do before you do it.
Now think about your own experience with AI tools. Wouldn’t you also like the answer up front, the structure to be predictable, and the assistant to tell you what it’s aboutto do?
The accessibility community isn’t asking for a different product. They’re askingfor a better product, sooner. Teams who listen ship the better versionfirst.
At Marino, wesee this pattern in our own work. The pattern is so consistent that we’ve started treating accessibility less like a compliance exercise and more like aroadmap exercise. When something is hard for a screen reader user, it’s almostalways harder than it needs to be for a tired user, a distracted user, a useron a poor connection, or a user encountering the product for the first time.Fix it once, and you fix it for all of them.
If you’re a service owner reading this and wondering whether accessibility is worth the investment beyond the legal floor - the European Accessibility Act now inf orce, WCAG 2.2, the Public Sector Bodies Directive - here’s the short version.
Accessibility work is more than a checklist; it’s the cheapest R&D budget you have. It acts as a mirror, reflecting the unexamined assumptions we bake into our interfaces. The 'standard' ways we expect users to see, hear, and interact.By surfacing real friction from real users, it cuts through the hand-waving of'stickiness' metrics to show you what’s truly broken. When you solve for those at the margins, you raise the floor for everyone.
In anAI-saturated market, the differentiator won’t be whether your product hasAI. Everyone’s product will have AI. The differentiator will be whether your product is calm, predictable, fair, and actually usable when someone is tired,stressed, or reading it through a screen reader at 7am on the bus into work.Accessibility is how you get there.
I wish I could say the industry has learned this lesson, but we aren't there yet. Many teamsstill treat accessibility as a 'nice-to-have' task to be tacked on at the endof a project. As a result, we’re seeing a new wave of AI tools hit the marketthat simply don't work for a huge portion of the population.
But the curb cut in Berkeley wasn’t built in a day either, and it’s hard to imagine a worldwithout it now. The accessibility work happening on your product right now isnot a tax. It’s a preview. Whatever your most accessibility-conscious users areasking for today, your average user is going to ask for by the end of nextyear.
Build for t hem first. Everyone else will catch up.
Have a project in mind, or just starting to think one through? We’re good at both.
Get in touch
