The Frontier Nobody Is Calling a Frontier

A few months ago I sat down with a vendor proposal for the platform I'm building. Dense, technical, priced. The kind of document that used to end me — not because I couldn't understand it, but because I'd read the same paragraph four times, lose the thread, and decide I'd deal with it later. Later never comes. That's how people like me end up signing things.

This time I read it, understood it, found the wrong part, and pushed back in writing the same day. They'd scoped roughly a hundred hours for work that realistically needed closer to two thousand.

Nothing about my brain changed between those two versions of that afternoon. I'm dyslexic, I show every sign of ADHD, and I got neither the word nor the help for thirty-five years. What changed is that a tool finally met me halfway.

I've told my own story in this magazine already. I don't want to tell it again. I want to talk about what I think is actually happening — because I don't think most people have noticed the size of it.

Every assistive tool before this one had a price of admission

Think about what accommodation has always required.

First, a diagnosis. Which requires knowing something is wrong, having someone to ask, being able to pay an assessor, and living somewhere assessors exist. Then a specific tool for a specific deficit — a screen reader, text-to-speech, a spellchecker, a note-taker. Then learning that tool. Then remembering to use it. Then, usually, a letter from an institution granting permission to use it.

Every one of those steps is a filter. And they compound in the same direction: toward people with money, in wealthy countries, inside institutions that have an accommodations office.

What's arrived in the last three years is categorically different, and the difference is not that the models are smart. It's the input.

These systems accept whatever shape your thinking comes in. Fragments. Out of order. Middle first. Spoken badly. Misspelled. Three ideas tangled together. They take that and hand back something structured — and then you can look at it and know instantly whether it's right, because recognising is easy for brains like mine even when producing in order is not.

That's the whole mechanism. Not intelligence. Tolerance of messy input. No prior technology in the history of assistive tech has been tolerant of messy input, because software has always required you to already be organised in order to get help being organised.

The number, and it isn't a national number

Here is what that opens up.

A global systematic review and meta-analysis in the Journal of Global Health put symptomatic adult ADHD at 6.76% of the world's adults — roughly 366 million people as of 2020. Persistent adult ADHD, the narrower definition, still came to about 140 million. The same analysis found prevalence running higher in low- and middle-income countries than in high-income ones, which is the reverse of where the diagnostic infrastructure sits.

Dyslexia is harder to count globally because diagnosis is so uneven, but the International Dyslexia Association's estimate is 15 to 20 percent of the population, and other bodies put it lower. Whatever the true figure, the diagnosed population is a small fraction of the actual one.

And then there is the number that reframes all of it. UNESCO counts 739 million adults who cannot read and write — down from 754 million the year before, but still. Nearly 77% of them live in sub-Saharan Africa and Central and Southern Asia. Two-thirds are women.

Those groups overlap, and they overlap in a way nobody planned. A person who never learned to read, a person whose dyslexia was never named, and a person whose attention has never once cooperated with a dense document are all locked out by the same door: the world runs on written text delivered at a pace and a density they cannot process.

That's not 15 million people in one country. That's on the order of a billion people, most of whom will never receive a diagnosis, and — this is the part that matters — no longer need one.

Removing the gate is the innovation

I want to be precise about the claim, because it's easy to overstate.

AI has not cured anything. What it has done is decouple help from diagnosis.

You do not need an assessment to ask a machine to say the paragraph again more simply. You do not need an accommodations letter. You do not need a specialist, a referral, a waiting list, or a school that has heard of your condition. You do not need to identify as anything at all. You ask, in your own words, badly, and you get something back you can use.

For someone in the US, that removes an assessment most people can't afford and a wait most can't sit through. For someone in a country with almost no diagnostic capacity for adult neurodevelopmental conditions, it removes an infrastructure that was never going to be built in their lifetime.

And the tool arrives in their language. Assistive technology has been overwhelmingly English-first for its entire history — every screen reader, every literacy product, every study aid, built for the market that could pay. These systems are not. That's not a small footnote. It is the difference between a technology that serves the diagnosed minority in rich countries and one that meets a person where they actually are.

I'm aware of how this sounds coming from someone who runs a company in this space. So test it against the thing you can check: the barrier to entry has never in history been this low, and the barrier to entry is what has always decided who gets help.

What I've seen, in the smallest possible sample

I drove rideshare through most of the build, and I asked passengers where they struggled.

Nurses. Law students. People halfway through degrees, all of them convinced their problem was discipline. One ride had four students in the car, and by the end of it they had designed most of my product: simplify text to a level I can actually process, help me write the email when the words come out sporadically, tell me what's next instead of showing me everything, remind me to eat.

What I hear now, from them and from the people on my waitlist, isn't that it makes me faster. It's some version of I can finally show people what I actually meant. That's not a productivity claim. That's a participation claim. It's the difference between working and being seen to be capable of working — and for most neurodivergent adults, the second one is what has actually cost them their careers.

Multiply that sentence by several hundred million, and you have the thing I think we are all standing in the middle of and not naming.

What it doesn't do

It isn't a cure. I still lose names within seconds. Words still come out wrong. During the build, I hit a burnout that lasted almost two months, and hyperfocus still eats whole days.

It also isn't free or universal. It needs a device, a connection, and enough digital literacy to start. That last one is a real gate, and it isn't going away by itself.

The translation problem got solved. The tax did not. But the tax was never what kept people out of rooms — the translation was.

Where the frontier actually is

The general-purpose tools got me most of the way, and further than the previous thirty-five years combined.

The last stretch is where people still fall off, and it's small and stupid: go open a different app for that. Hold your place, learn a new interface, come back. That handoff is where tasks die — not because it's hard, but because it's one more thing, and one more thing is the whole ballgame when you're already at capacity.

That's the gap I'm building into, and it's the gap I'd point any builder toward. The models are already smarter than the problem. What's missing is everything around them: continuity, removing the switching, and a default that doesn't assume the user arrives organised.

I want to be careful here, because the argument I've just made is about a technology and not about my company. UnSquirrel launches in the United States, and it stays there until every feature works without friction. Not because the need is American — the numbers above say plainly that it isn't — but because shipping something friction-heavy to people who are already at capacity is worse than not shipping to them at all. Friction is the entire problem I'm trying to solve. I'm not going to export a version of it.

International comes after. It comes when the thing actually works.

The chair

I walked out of high school at sixteen, in the middle of advanced chemistry, and it took thirty-five years to understand what had happened to me.

Somebody is sixteen right now in that same chair, and somebody else is sixteen in a classroom four hundred miles from the nearest assessor. Historically, those were two entirely different stories with two entirely different endings, and the difference between them was never ability. It was who could afford to be told what was wrong.

Neither of them is waiting on my company. That's the point. The gate is already down. What's left is for the rest of us to build the things that walk through it.