Synthesize

An AI-powered form building and feedback analysis platform. Create surveys, collect responses via an adaptive AI chatbot, and get actionable business insights automatically synthesized from your data.

Features

• AI Form Builder — Generate survey questions from a business context and information goals
• Manual Form Builder — Create custom forms with full control over questions
• AI Chatbot Survey — Respondents answer through a conversational GPT-4o agent that adapts follow-up questions based on their answers
• Automated Analysis — Responses are processed incrementally using GPT-4o to extract sentiment (positive/negative) and actionable suggestions
• Dashboard — View aggregated sentiments, statistics, and prioritized actionable insights per form
• Project Organisation — Group forms under projects for multi-study workflows

Tech Stack

Layer	Technologies
Frontend	Next.js 14, TypeScript, Tailwind CSS, Radix UI, Redux Toolkit
Backend	Python, FastAPI, Uvicorn
AI	OpenAI GPT-4o
Database	Firebase Firestore
Auth	Firebase Authentication (email/password, Google, GitHub)

Project Structure

Synthesize/
├── backend/
│   ├── main.py                        # FastAPI app & API routes
│   ├── requirements.txt
│   └── services/
│       ├── OpenAiService.py           # GPT-4o: analysis, merging, chatbot
│       ├── FirebaseService.py         # Firestore read/write
│       ├── ChatBotService.py          # Adaptive survey chatbot logic
│       └── utils/
│           └── system_prompt_templates.py
└── frontend/
    ├── app/
    │   ├── auth/                      # Login & signup pages
    │   └── platform/
    │       ├── (withNavbar)/
    │       │   ├── homepage/          # Home dashboard
    │       │   ├── projects/          # Project management
    │       │   ├── form/              # Form creation (AI & manual)
    │       │   ├── formsCreated/      # View forms & responses
    │       │   └── dashboard/        # Insights dashboard
    │       └── (withoutNavbar)/
    │           └── chat/              # AI chatbot survey interface
    ├── components/                    # Reusable UI components
    ├── database/                      # Firebase CRUD functions
    ├── lib/
    │   ├── firebase/                  # Firebase config & auth helpers
    │   └── types.ts                   # Shared TypeScript types
    └── redux/                         # Auth state management

Getting Started

Prerequisites

• Node.js 18+
• Python 3.11+
• Firebase project (Firestore enabled)
• OpenAI API key

Backend Setup

cd backend
python -m venv venv
source venv/bin/activate      # Windows: venv\Scripts\activate
pip install -r requirements.txt

Create backend/.env from the example:

cp .env.example .env

OPENAI_API_KEY="your_openai_api_key"
DATABASE_URL="your_firebase_database_url"
SECRET_KEY="your_secret_key"

Place your Firebase service account credentials at backend/firebase_credentials.json.

Start the server:

uvicorn main:app --reload
# API available at http://localhost:8000

Frontend Setup

cd frontend
npm install

Create frontend/.env.local from the example:

cp .env.local.example .env.local

NEXT_PUBLIC_NODE_ENV="local"
NEXT_PUBLIC_MOCK_USER_EMAIL="your_email@example.com"

Start the dev server:

npm run dev
# App available at http://localhost:3000

API Reference

Method	Endpoint	Description
GET	/	Health check
GET	/get_form_analysis/{formId}?secret=...	Trigger or retrieve AI analysis for a form
POST	/chat/{projectId}/{formId}	Send a message to the AI survey chatbot

The /get_form_analysis endpoint uses a 10-minute cache — if an analysis was run within the last 10 minutes it returns the stored result without calling OpenAI again.

Running Tests

cd frontend
npm test

Tests use Jest + React Testing Library. Coverage reports are written to frontend/coverage/.

Deployment

The frontend is deployed on Vercel. The backend CORS policy allows:

• http://localhost:3000 (local dev)
• https://synthesize-two.vercel.app (production)
• https://synthesize-git-develop-jun-yus-projects.vercel.app (preview)