Pictro

Pictro is the project I'm proudest of, and the closest thing I have to an example of the software I want to make.

What it does

A psychologist takes a real activity and breaks it into steps. Each step is a pictogram and a few words. The child follows along on a phone while doing the activity in real life, and marks each step when it's done.

A routine is a real activity, one small step at a time

Each step is a picture and a few words. The child always sees what is done, what is happening now and what comes later.

1. Wake upLaterNowDone!
2. Brush teethLaterNowDone!
3. Get dressedLaterNowDone!
4. EatLaterNowDone!
5. Pack bagLaterNowDone!

The psychologist builds every routine for one child, on the web. The child gets only what was prepared for them, in the order the psychologist chose.

Designing for autistic children

I started from published guidelines on designing for people on the autism spectrum. A few of them ended up shaping almost every screen.

• Every idea gets a picture and words, side by side.
• One thing at a time, with very little on the screen.
• Nothing moves or changes where it wasn't expected.
• Feedback is positive and short.
• If the device asks for less motion, the app stays still.

Color never carries the meaning on its own. A step says "Now", "Done!" or "Later" in words, with an icon beside it.

Why it works like a game

My thesis started from a problem with the tools that already existed. Most of them gave a child no reason to come back. When the child loses interest, the routine stops, and the treatment loses what it was building on.

So finishing a routine earns something. In the first version that meant stars that filled in as each step was done, a reward chosen by the psychologist, and an album to keep the rewards in.

The version I'm building now has three things to earn.

Three things to earn, each with one job

A finished routine pays out all three. Steps on their own pay nothing.

• CoinsBuy clothes for the cat and furniture for its room.
• StarsTrade for a real reward the psychologist set up: a toy, an outing, time doing something they love.
• StickersOne surprise at the end of every routine, to stick in an album.

I chose a few rules on purpose.

• The reward comes when the whole routine is finished. Navigating through the steps is not finishing.
• Nothing is ever taken away. There are no penalties and no lost points.
• The days a psychologist picks are suggestions. A child can do a routine on any day.
• A routine that reached its daily limit says it's done for today. It never says it's locked.
• The psychologist can turn off a theme for a child who doesn't like it or is afraid of it.

What testing taught me

I tested the first version with ten people, among them psychologists who work with autism and people who care for someone on the spectrum.

92.5 on the System Usability Scale

Ten participants tried the first version. The average score for software is 68.

92.5out of 100

average, 68

0100

A comment that came up more than once, in my translation, was that the stars "help a lot with immediate encouragement for the patient."

They also asked for things I hadn't built. They wanted rewards that come as a surprise, which is how many clinics already do it. They wanted a way to trade what the child earns for something real. They wanted routines that work without internet.

Those requests became the center of the rebuild. The sticker is a surprise. Stars are traded for real rewards. Working offline is planned for a later step.

The cat

The child has a companion, a cat that lives in a room they decorate. It breathes, swishes its tail and blinks. Tap it and it smiles, waves and hops, and a heart floats up. With reduced motion turned on, it stays still and only its face changes.

It's the same cat that has been following you down this page. You can pat it.

How it's built

The thesis version was a React Native app for the child and a React app for the psychologist, on a Node.js backend running as serverless functions.

It had to cost almost nothing to run, so I chose Turso, UploadThing and Vercel.

I'm rebuilding it now as one codebase, with Expo for the child's app, Next.js for the psychologist and Convex behind both. The rules live in the backend, so the two apps can't disagree about them.

The plan for AI is narrow. It drafts a routine and its pictograms from a sentence the psychologist writes. The psychologist reviews every draft before a child sees it, and nothing from the patient's record goes to the model.

My part

Pictro started as my bachelor's thesis in Software Engineering at UDESC. I built it together with professionals in psychology and defended it in November 2024 with the top grade.

I'm a founder of the business behind it, and I led every stage, from the interviews to the design, the code and the tests.

Where it stands

Pictro isn't public yet. The screens are being rebuilt and tested with sample data. I still believe in it, and I intend to launch it.