I've been in disability support for fifteen years.

For most of those years, I was doing two things at once. The job I was paid for, and the job I was actually trying to figure out.

The paid job was support work. Showing up. Helping people get through their week. The usual.

The other job was harder to explain. I kept trying to build something. A system that could hold what a neurodivergent brain couldn't. An externalised memory. Something that captured, organised, and remembered so the person didn't have to.

I didn't have that language for it yet. I just knew the pattern. Every client I worked with had the same problem: too much in their head, no reliable place to put it, and every app that promised to help required the exact thing they couldn't do.


The first real attempt was Trello.

If you haven't used it, Trello is a digital board with columns and cards. You drag things from one column to the next. It's clean. It's visual. It looks like progress.

I set up boards for my clients. Plans. Tasks. Invoices. Goals. Everything in columns, colour-coded, labelled, linked. Day one, it looked great.

Week three, nobody was opening it.

Not the clients. Not me.

The cards sat there. The plans went stale. The invoices stayed in the same column. The board became a snapshot of good intentions from three weeks ago.


Here's what I didn't understand at first.

I thought the problem was the tool. Trello wasn't right. The layout was wrong. The columns needed renaming. Maybe a different app would stick.

So I'd try something else. A different tool. A better setup. A cleaner layout. And every time, the same thing happened.

Day one: perfect.

Week three: abandoned.

It took me longer than I'd like to admit to realise the tool was never the problem.


The problem was friction.

Every system I built required effort to maintain. Not a lot of effort. Just a little. Open the app. Move the card. Update the status. Check the board. Five minutes.

Five minutes is nothing if your executive function works.

Five minutes is everything if it doesn't.

ADHD doesn't take away your ability to use a system. It takes away your ability to maintain one. The difference matters. On a good day, you can set up anything. On a normal day, you can't sustain it. On a bad day, you can't open it.

I was building systems that worked on good days and disappeared on normal ones. The tool asked the person to do the thing the person couldn't do. That's not a user problem. That's a design problem.


At the same time, I was teaching my clients something else.

AI.

Not the full technical picture. Just the practical part. How to use it to think through problems. How to ask it questions when your brain wouldn't cooperate. How to take the mess in your head and turn it into something structured.

I could see the power. A neurodivergent person could talk for two minutes and an AI could organise what they said into something clear. That was real. That was useful.

But it disappeared.

Every conversation vanished when the session ended. There was no memory. No accumulation. No system underneath. The AI was clever, but it was stateless. You'd have the same breakthrough conversation three times because nothing carried over.


So I had two halves of an answer and neither worked alone.

Structure without intelligence was Trello. Beautiful boards that went stale because nobody maintained them.

Intelligence without structure was AI. Brilliant conversations that evaporated because nothing was captured.

The person captured into Trello, but the system couldn't think. The person talked to AI, but the system couldn't remember.

I needed them fused. One system where the person captures, the AI organises, and nobody has to maintain anything. The structure maintains itself because the intelligence maintains it.


I didn't have the language for this yet. I hadn't written the seven principles. I hadn't read the research. I didn't know that Russell Barkley had spent decades arguing the same thing from a clinical perspective: if the internal function is impaired, re-externalise it. Put the memory, the cues, the decision structures back into the environment. Don't train the brain. Support it at the point of use.

I just knew the pattern from watching it fail.

Set up a system. Watch it work. Watch it stop working. Watch the client feel bad about it. Watch the client blame themselves for something the system did to them.

That last part was the one that got to me.

Every abandoned board, every stale plan, every app that stopped being opened. The client didn't think "this tool failed me." They thought "I failed again."

The tool created guilt. The tool that was supposed to help was generating shame every time the person couldn't maintain it.


That's what I carried into the next thing I built.

Not just "it needs to work." Not just "it needs AI." Not just "it needs structure."

It needs to be the kind of system where nothing breaks if you rest.

Where coming back after a week away doesn't mean facing a wall of things you didn't do. Where the system meets you where you are, not where you planned to be.

The person captures. The system organises. The system maintains itself. And if the person disappears for a while, the system is still there when they come back. No guilt. No backlog. No score.

I knew what the system needed to be. I just hadn't found the tools that could do it yet.


Jamie Robinson is the founder of Neuro Support, a specialist neurodivergent support service based in Christchurch, New Zealand. Previous: How Blizzard Broke My Healer and I Accidentally Started a Business.