Clario

Clario helps patients understand hospital bills in plain English, spot overcharges, and generate dispute letters.

---

What it does

1. Patient uploads a medical bill (PDF or photo)
2. OCR extracts the text from the document
3. AI analyses every charge and flags suspicious items
4. Patient gets a plain-English breakdown and can download a dispute letter, phone script, or complaint letter

---

Tech stack

Layer	Tool
Framework	Next.js 15 (App Router) + React 19
Language	TypeScript (strict)
Styling	Tailwind CSS v3 + Framer Motion
AI	Groq (llama-3.3-70b) or Gemini 2.0 Flash
OCR	Vision model (Groq/Gemini) for images, pdf-parse for PDFs
Database	Supabase (Postgres + Auth + Storage)
Deployment	Vercel

---

Getting started

# 1. Clone the repo
git clone https://github.com/PARTHDEVX2904/clario-app.git
cd clario-app

# 2. Install dependencies
npm install

# 3. Set up environment variables
cp .env.local.example .env.local

# 4. Start the dev server
npm run dev

Open http://localhost:3000 in your browser.

---

Environment variables

Variable	Required	Notes
NEXT_PUBLIC_SUPABASE_URL	Yes	From Supabase project settings
NEXT_PUBLIC_SUPABASE_ANON_KEY	Yes	From Supabase project settings
SUPABASE_SERVICE_ROLE_KEY	Yes	Server-only — never expose to client
GROQ_API_KEY	If using Groq
GEMINI_API_KEY	If using Gemini
AI_PROVIDER	Yes	mock, groq, or gemini
OCR_PROVIDER	Yes	mock or tesseract

---

Project structure

app/
  (marketing)/     # Public pages — landing, privacy
  (app)/           # Authenticated pages — dashboard, upload, analysis
  api/             # Server-only API routes

components/
  ui/              # Reusable primitives — Button, Card, Badge
  layout/          # Sidebar, Header, AppHeader
  dashboard/       # Stats cards, recent analysis list
  upload/          # Multi-step bill upload form
  analysis/        # Analysis result display

lib/
  ai/              # AI provider adapters (groq, gemini, mock)
  ocr/             # OCR adapters + image preprocessing
  supabase/        # DB clients — browser + server
  utils/           # cn(), formatCurrency(), formatDateRelative()

types/             # Shared TypeScript types + Supabase DB types

---

Database setup

Run lib/supabase/schema.sql in your Supabase Dashboard → SQL Editor.

Tables created (all with Row Level Security):

• cases — patient episode context
• documents — uploaded bill files + OCR output
• bill_analyses — AI analysis results per case
• bill_line_items — individual charge line items
• generated_outputs — dispute letters and scripts
• profiles — public mirror of auth.users, auto-created on signup

---

Bill processing pipeline

Upload → Preprocess image (sharp) → Vision OCR → Clean text → LLM analysis → Validate JSON → Save to DB

Charge flag types:

Flag	Meaning
valid	Charge is consistent with the care reported
review_needed	Worth questioning — ask for itemised breakdown
possibly_overcharged	Significantly above expected norms

All flags are informational only — not legal or medical advice.

---

Scripts

npm run dev       # Dev server
npm run build     # Production build
npm run lint      # ESLint

---

Deployment

1. Push to GitHub
2. Import project in Vercel
3. Add all environment variables in Vercel project settings
4. Deploy

---

Disclaimer

Clario provides informational support only. It is not legal or medical advice. Always consult a certified patient advocate or healthcare attorney for definitive guidance.