Vrataski — human-in-the-loop / accessibility pitch

Living external-facing pitch that melds the human-in-the-loop design ethos with the personal accessibility (executive-dysfunction) framing. First-person, deliberately jargon-free, written for a general or AI-skeptical audience. Drafted 2026-06-13.

Most AI right now is built to run on its own — you hand it a task and it goes off and does the whole thing start to finish, and you're left to either trust the output or untangle it. What I've been building is the opposite by design. It takes the shape of those autonomous systems but puts a person back into every step: the human stays in the loop the whole way through, and the AI never decides or acts on its own. There's no point where it quietly takes the wheel — no autonomous process running that I'm merely supervising. I'm in it at every turn.

That changes what the AI is. It's a thinking partner and a set of hands — never a replacement for the judgment. It proposes; I dispose. When it checks or runs something, the result is evidence I weigh, never a decision it makes for me. And the whole thing is deliberately calm: it offers and suggests rather than pushing or pressuring, because a tool that nags or demands is one I'd come to resent and avoid.

For me, that design isn't an abstraction — it's what makes it an accessibility tool. I'm autistic and have ADHD, and the hard part of getting things done was never the thinking — it's the executing. With software, the problem-solving and the design, the actual engineering, is the part I'm good at and genuinely enjoy. What drains me is the mechanical grind: the literal typing, the syntax, getting it out of my head and onto the screen. That's executive dysfunction — task initiation, working memory, shifting gears — not a limit on what I'm capable of. Because the AI stays in the loop and I stay in charge, it can take over exactly that part: I do the reasoning and make every call, and it handles the costly execution, without ever taking the work away from me.

Because the wall is in the doing, not the subject, it isn't only for code. Software is the first-class case, but the same thing helps me execute something like planning a trip — anywhere the barrier is the legwork, not the decisions. And it holds the thread for me: I can step away for twenty minutes or a year, come back, and it reorients me, because the work and its whole history live in plain documents and version control rather than anyone's memory — including a running log of why each decision was made. So when I come back to a choice I made but can't remember the reasoning behind, the reasoning is still there to read. For a brain that drops threads and bounces between things, that's the difference between resuming and starting over.

The interaction itself costs less, too: it doesn't require the masking I'd have to do with a person, and that same softness — offering, never pushing — is what makes it workable for someone who's demand-avoidant.

So the two halves are really one idea: keeping the human at the center isn't just a philosophy, it's the exact thing that lets the tool carry my executive-function load without carrying off my agency with it. It's not AI replacing me. It's an accessibility tool that closes the gap between what I can think and what I can actually build — and keeps my place for me when I can't.