Government Forms Made Simple, Fast, and Flawless.
Formitra AI is a next-generation automation ecosystem designed to eliminate the friction, errors, and sheer time sink of applying for Indian government services. By combining localized edge-compute, cloud Vision AI, and a specialized Chrome Automation Extension, Formitra transforms grueling, hour-long bureaucratic chores into a seamless 3-minute click-through experience.
- Objective 1: Eliminate Manual Data Entry
- Achievement: Built an intelligent extraction pipeline that reads unstructured ID documents (Aadhaar, PAN) and maps them to strict government schemas almost instantaneously using LLM Vision mapping.
- Objective 2: Enforce Zero-Leak Privacy Infrastructure
- Achievement: Implemented Local WebAssembly (WASM) image processing. Features like background removal and passport photo standardization execute entirely within the user's browser CPU—preventing sensitive biometric photos from ever leaving the local machine.
- Objective 3: Bridge the Legacy Gap
- Achievement: Engineered a cross-origin Chrome Extension bridge. It listens for payload dispatches from the modern React app and automatically DOM-injects the data sequentially into legacy
.gov.insites, requiring zero un-authorized API access.
- Achievement: Engineered a cross-origin Chrome Extension bridge. It listens for payload dispatches from the modern React app and automatically DOM-injects the data sequentially into legacy
- Objective 4: Enterprise-Grade B2B Aesthetics
- Achievement: Designed a deeply interactive, responsive UI powered by native Tailwind CSS v4, featuring glassmorphic overlays, advanced hover physics, and strict, accessible design tokens.
The ecosystem relies on three distinct layers working in harmony:
The gorgeous, user-facing portal where citizens select services, upload documents, and verify AI-extracted data.
- Core Tech: React 19, Vite 7, React Router DOM.
- Styling & UI: Tailwind CSS v4, Framer Motion, Lucide React icons.
- Key Innovation: Utilizes local Web Workers via
@imgly/background-removal. If a user uploads a casual selfie, the React app instantly strips the background and converts it to a government-compliant white-background passport photo locally.
A lightweight, high-performance REST API acting as the intelligence broker.
- Core Tech: Node.js, Express, Multer.
- AI Integration: Google GenAI SDK (
gemini-1.5-pro). - Key Innovation: Zero-disk-storage pipelines. User documents are held strictly in RAM (
multer.memoryStorage()), streamed to the Gemini Vision API for semantic extraction, mapped into the requiredformitra_form_dataJSON schema, and immediately purged from memory.
The robotic automation bridge navigating legacy .gov domains.
- Core Tech: Vanilla JavaScript, Chrome Manifest V3 APIs.
- Key Innovation: Uses an asynchronous
content.jslistener andbackground.jsservice worker. It detects "Submit Application" broadcasts fromlocalhost:5173, navigates to the target government URL, and executes localized scripts to rapidly type, click, and populate the applicant's data flawlessly.
- Service Selection: User visits the robust Formitra landing page and selects a service pipeline (e.g., Passport Services > Tamil Nadu).
- Smart Document Ingestion:
- User uploads supporting documents (Aadhaar, PAN, Photo) into the "AI Intake Enabled" drop-zone.
- The local React WASM engine optionally cleans and standardizes the photographs.
- Vision AI Extraction:
- Encrypted files are sent to the Express server.
- Gemini 1.5 Pro visually scans the documents, extracting deeply nested and unstructured data (DOB, full formatted address, IDs) into strict application schemas.
- Verification & Validation: The user reviews the auto-filled form in the React application. The UI enforces strict regex validation to ensure data won't be rejected by the government site.
- Extension Hand-off (DOM Injection):
- Upon confirmation, the React app stores the payload securely in local storage and triggers the Chrome Extension.
- The extension hijacks the operation, opens the corresponding
passportindia.gov.inURL, and sequentially injects the payload into the archaic DOM forms in seconds.
Frontend Client:
react^19.0.0 (DOM Architecture)@tailwindcss/postcss&tailwindcss^4.1.18 (Next-Gen Styling Engine)framer-motion(Fluid Physics Animations)@imgly/background-removal(Local Machine Learning execution)browser-image-compression(Pre-flight optimization)
Backend API:
express(Network Routing)multer(Multipart/form-data RAM streams)@google/genai(LLM Vision Models)dotenv(Environment management)
Browser Automation:
- Chrome Extension APIs (Manifest V3, Service Workers, Script Injection)
1. Boot the Server:
cd server
npm install
# Ensure .env contains GEMINI_API_KEY=your_key_here
node server.js
# API running on Port 30002. Boot the Client:
cd client
npm install
npm run dev
# Portal running on http://localhost:51733. Install the Brain (Extension):
- Open Google Chrome > Navigate to
chrome://extensions/ - Enable Developer Mode (toggle top right).
- Click Load Unpacked and select the
/extensiondirectory inside this project.
Demo Flow:
- Open
http://localhost:5173. - Click "Get Started" and select Passport Services.
- Upload an ID card. Watch the AI Extraction spinner.
- Review the perfectly auto-filled form.
- Click "Auto Fill via Formitra Extension" and watch the magic happen on the government portal.
Formitra AI — Architected for speed, built for precision, and designed to redefine Indian citizen-government interactions.