The story starts with a cliff.
At 22, structured support for autistic people in the US falls off a ledge. School services end, and what replaces them is a handful of coaches and programmes who work with adults — if you live near one, can afford the hours, and can wait out the waitlist.
So the default becomes the only free option: practicing on real life. Every interview is a first take with income on the line. Every friendship is maintained by rules nobody will state. Every mistake is public, and the feedback is rejection. It would be hard to design a worse learning environment on purpose.
Subtext exists because that rehearsal gap is fixable now. The same AI that can hold a natural conversation can play a hiring manager at 11pm the night before your interview — patiently, privately, for the twelfth time in a row — and then tell you, in plain language, what the other person's signals meant. Not therapy. Not treatment. Practice.
We're building it the only way it should be built:
Nothing about us without us — as an operating rule, not a slogan. A paid autistic advisory board has held veto power over framing and feedback design since before the first line of code. Beta co-designers are compensated, credited, and keep free access for life. The feedback engine's "never" list (no eye-contact drills, no normalcy scores) is enforced in the product's architecture, not just its marketing. And we build in the open — when the community tells us something reads wrong, we change it and show our work.
We won't pretend to a backstory we don't have. We're a small team that read the research, listened in the places autistic adults already talk — and kept finding the same sentence: "I know the rules. I've just never had anywhere safe to practice them." That's the product.



