InspectAI

A premier inspection intelligence platform designed for property management, restoration, and real estate professionals. InspectAI leverages cutting-edge Gemini vision models to automatically analyze walkthrough videos, map property defects, identify structural anomalies, and calculate localized cost estimations in real time.

---

🏗️ Architecture

InspectAI is built using a modern Decoupled Architecture to optimize performance, cost, and user experience:

• Frontend (Vercel): A fast, React 19 SPA. It loads instantly and remains always-awake.
• Backend (Render): An Express-based monolithic Node.js API container designed to handle heavy multi-modal AI processing tasks and secure uploads.
• Storage (Cloudflare R2): Large videos are uploaded directly from the browser using AWS S3-compatible presigned URLs, bypassing the server to maximize bandwidth.
• Database & Auth (Firebase): Google OAuth authentication and Firestore database storing user preferences and inspections.

---

🛠️ Step-by-Step Deployment Guide

Deploying InspectAI involves configuring four key components: Firebase, Cloudflare R2, the Render Backend, and the Vercel Frontend.

1. Firebase Configuration (Auth & Database)

1. Go to the Firebase Console.
2. Create a new Firebase project.
3. Enable Authentication:
   • Navigate to Authentication > Sign-in method.
   • Enable the Google provider.
4. Create a Firestore Database:
   • Navigate to Firestore Database > Create database.
   • Set up rules inside firestore.rules (provided in the repository) to restrict read/write access to authenticated owners.
5. Whitelist Frontend Domain (Crucial for Mobile/Desktop OAuth):
   • Navigate to Authentication > Settings > Authorized domains.
   • Add your local domain: localhost
   • Add your deployed Vercel domain: inspect-ai-ruddy.vercel.app (replace with your production URL).

---

2. Cloudflare R2 Setup (Video Storage)

Cloudflare R2 acts as the high-capacity, low-latency video vault.

1. Create a bucket in your Cloudflare R2 dashboard (e.g., inspect-ai).
2. Generate an API token with Edit permissions. Save the Access Key ID and Secret Access Key.
3. Enable Public R2.dev URL (or bind a custom domain) in the bucket settings.
4. CORS Policy Setup:
   • Go to Bucket Settings > CORS Policy > Edit.
   • Paste the following JSON to whitelist your development, Render, and Vercel domains:

   [
     {
       "AllowedOrigins": [
         "http://localhost:5173",
         "https://inspectai-e49p.onrender.com",
         "https://inspect-ai-ruddy.vercel.app"
       ],
       "AllowedMethods": [
         "GET",
         "PUT",
         "POST",
         "HEAD"
       ],
       "AllowedHeaders": [
         "*"
       ]
     }
   ]

---

3. Backend Deployment (Render.com)

The backend container spins down on Render's free tier, but the frontend features an automatic background warm-up loading bar to manage the transition smoothly.

1. Create a Web Service on Render.
2. Connect your GitHub repository.
3. Configure the build parameters:
   • Build Command: npm install && npm run build
   • Start Command: npm start
4. Add the following Environment Variables:
   • GEMINI_API_KEY: Your primary Gemini API key (acts as the server fallback).
   • R2_ENDPOINT: Your Cloudflare S3 endpoint (https://<account_id>.r2.cloudflarestorage.com).
   • R2_ACCESS_KEY_ID: Your Cloudflare Access Key.
   • R2_SECRET_ACCESS_KEY: Your Cloudflare Secret Key.
   • R2_BUCKET_NAME: The name of your R2 bucket.
   • R2_PUBLIC_URL: Your bucket's public domain URL.

---

4. Frontend Deployment (Vercel)

The frontend is completely serverless, ensuring 100% uptime with zero sleep timeouts.

1. Link your GitHub repository in the Vercel Dashboard.
2. Vercel automatically detects the Vite configuration.
3. Set the following Environment Variables:
   • VITE_API_URL: Your Render backend URL (https://inspectai-e49p.onrender.com).
   • VITE_FIREBASE_API_KEY: Firebase API Key.
   • VITE_FIREBASE_AUTH_DOMAIN: Firebase Auth Domain.
   • VITE_FIREBASE_PROJECT_ID: Firebase Project ID.
   • VITE_FIREBASE_STORAGE_BUCKET: Firebase Storage Bucket.
   • VITE_FIREBASE_MESSAGING_SENDER_ID: Firebase Sender ID.
   • VITE_FIREBASE_APP_ID: Firebase App ID.
4. Deploy!

---

🚀 How to Run Locally

1. Clone the Repository:

   git clone https://github.com/lead-with-data/InspectAI.git
   cd InspectAI

2. Install Dependencies:

   npm install

3. Setup Environment: Create a .env file in the root directory and populate it with the variables from .env.example.
4. Run Development Servers:
   • Start the API backend: npm run server (runs on http://localhost:3000)
   • Start the React frontend: npm run dev (runs on http://localhost:5173)
5. Open http://localhost:5173 to explore.