🌿 Inspiration

EaseFlow was inspired by a simple but important problem:

Many autistic children benefit from structured routines and clear, literal instructions — yet guardians often need to remain physically present to guide them through everyday tasks.

We wanted to build a system that creates a calm, structured bridge between guardian and child — enabling independence without removing support.

Instead of replacing guardians, EaseFlow supports them.

The idea is based on one principle:

Independence

Structure + Clarity + Predictability Independence=Structure+Clarity+Predictability

If tasks are broken down into manageable, step-by-step instructions, and presented in a sensory-friendly interface, children can perform them more confidently and independently.

🧠 What We Learned

Building EaseFlow taught us much more than just Flutter and Express.

We learned:

How to design for neurodiversity

How to reduce cognitive overload in UI

How to implement role-based authentication

How to structure Firestore collections effectively

How to verify Firebase tokens securely in Express

How to build a lightweight AI endpoint using a Flan-T5 model

How to structure a cross-platform Flutter application

We also learned that accessibility is not just about larger fonts — it is about:

Predictability

Minimalism

Emotional safety

Clear communication

🏗 How We Built It

EaseFlow consists of three major components:

1️⃣ Backend (Node.js + Express + Firebase Admin)

The backend handles:

User authentication (via Firebase)

Role-based logic (guardian/child)

Task creation and step management

Profile storage

Token verification middleware

AI text-generation endpoint

We implemented structured routes such as:

POST /auth/sign_up POST /tasks POST /tasks/:taskId/steps GET /tasks/child/:childUid POST /profile POST /ai/generate

The backend uses Firebase Admin SDK to:

Verify ID tokens

Read/write Firestore documents

Maintain secure role-based access

2️⃣ Mobile & Desktop App (Flutter)

The Flutter app supports:

Android

iOS

Windows

macOS

Linux

Web

The app is structured with:

Role-based navigation

Guardian dashboard

Child dashboard

Profile screens

Task step views

Simple theme switching

Children see:

Large buttons

Minimal screens

Clear task steps

Emoji/image-supported instructions

Guardians see:

Task management

Child linking

Profile customization

Monitoring capabilities

3️⃣ AI Service

We implemented a lightweight AI endpoint powered by a Flan-T5 model via @xenova/transformers.

It allows:

Simple prompt-based text generation

Future expansion for step simplification

Adaptive communication tone

Although optional, it demonstrates how AI can assist structured task rewriting in future versions.

⚙ Architecture Overview Flutter App → Express Backend → Firebase (Auth + Firestore) ↘ AI Service (Flan-T5)

Authentication flow:

User signs in via Firebase Auth

Flutter sends ID token to backend

Backend verifies token with Firebase Admin

Role-based access control is applied

🧩 Challenges We Faced 🔐 Token Verification & Security

Understanding how Firebase ID tokens work and correctly verifying them in Express middleware was one of the biggest backend challenges.

📱 Multi-Platform Flutter Support

Ensuring the UI remained consistent across:

Mobile

Windows

macOS

Linux

required careful layout decisions and responsive design.

🧠 Designing for Autism

The hardest challenge was not technical.

It was UX.

We had to constantly ask:

Is this screen too crowded?

Is this instruction too abstract?

Is this animation overstimulating?

We learned that good accessibility design often means removing features, not adding them.

🤖 AI Integration

Integrating a local transformer model required:

Managing model loading

Optimizing response generation

Handling structured output

💡 Why This Project Matters

Autistic children often thrive in structured environments. However, daily routines can become overwhelming without consistent guidance.

EaseFlow encourages independence while preserving connection.

Instead of replacing guardians, it extends their presence through structured digital support.

Our vision is to continue expanding the system with:

Adaptive AI rewriting of task steps

Now/Next/Later scheduling

Alert-based guardian notifications

Personalized communication styles

🚀 Future Improvements

Full AI-based step simplification

Visual schedule timeline

Push notifications for guardian alerts

Analytics for task completion patterns

Improved sensory settings (reduced animation mode, color themes)

❤️ Final Thought

EaseFlow is more than a task app.

It is a small step toward:

Independence

Structure

Confidence

Neurodiverse-friendly technology

One task at a time.

Built With

Share this project:

Updates

Submission history