Autism Visual Supports

First → Then. Print it, post it, trust it.

Your child is on the floor with a toy. In four minutes they need shoes on with their go-bag and they need to be in the car. We have to go now does not work, and saying it louder works less. What works is a picture — First shoes. Then swing. Two cards, in order, where the child can see them. That much is established practice, documented in the Vanderbilt Kennedy Center's TRIAD toolkit and backed by the National Autism Center's evidence review. What fails is the making: finding pictures the child recognises, cutting them to size, keeping every label to three words, laminating, adding Velcro. Forty minutes — and the morning you needed it was this morning. So most parents make two or three boards, ever, and then stop.

Autism Visual Supports is an OutSystems ODC application that asks three questions — which routine, what is hard about it, what your child actually wants afterwards — and returns a printable board, the words to say out loud, and a plan for when it goes wrong, in under two minutes. Three agents draft it, review it against the design rules, and reuse pictures the child already knows, grounded in the toolkit rather than invented. But a generator that stops at a pretty PDF has only solved the easy half. The toolkit is blunt about the other half: deliver the promised reward, "otherwise, your child may not trust the board the next time you use it." A board the child has stopped trusting is worse than no board. So Twilio texts you the script when the routine is due, and afterwards asks one question — did the reward happen? That single reply is the field the whole method rests on, and the one no parent would ever open an app to fill in, because at that moment they are at the front door holding a pair of shoes. The home screen tracks promises kept, not boards made.

OutSystems ODC · Agentic AI · Twilio · Build for the Future Hackathon 2026

Open source: https://github.com/aj-aronoff/Autism-Visual-Supports