VendorLite

Inspiration

As a creator who wanted to sell digital products online, I was frustrated by how many tools I needed just to get started. Shopify for the store, Mailchimp for email, Calendly for bookings, a CRM for customers, a blog for content - the list kept growing. Each tool had its own subscription, dashboard, and learning curve. By the time I got everything set up, my margins were gone and my motivation was drained.

I built VendorLite because creators deserve better. One platform. One dashboard. One subscription. Everything you need to sell digital products, courses, and services - without writing a single line of backend code or wrestling with plugin integrations.

What I Learned

This project taught me the power of building on a unified backend. Using Supabase as the backbone (Auth, Postgres, Storage, and Edge Functions) meant I could focus on features instead of infrastructure. I learned how to implement real-time SSE streaming for AI responses, how to structure a multi-role authentication system with 2FA, and how to integrate Stripe payments securely through webhooks.

I also learned that AI isn't a feature you bolt on - it's a workflow you embed. The AI tools (Web Search, Assistant, and Web Reader) aren't separate pages that users open and close. They're part of the daily creator workflow, accessible from the same sidebar as Orders and Analytics.

How I Built It

Frontend: React 18 + Vite + TypeScript + Tailwind CSS + shadcn/ui. The UI is fully responsive with dark mode support. Marketing pages use Intersection Observer for smooth scroll-triggered animations.

Backend: Supabase handles everything - Auth (email/password + TOTP 2FA), Postgres database with Row-Level Security, Storage for product images, and 14+ Edge Functions for server-side operations.

Payments: Stripe Checkout for secure payment processing with full webhook verification.

AI: Three AI tools powered by Gemini 2.5 Flash, all proxied through Supabase Edge Functions using SSE streaming for instant, typewriter-style responses.

Database: Relational schema with foreign keys, enums, and triggers. Tables include products, orders, customers, bookings, invoices, affiliates, blog posts, login attempts, and user profiles.

Challenges I Faced

The biggest challenge was implementing real-time AI streaming. Getting SSE to work through Supabase Edge Functions required careful handling of response headers and body streaming. Parsing Gemini's response chunks while maintaining state for sources and citations added complexity.

Another challenge was building a multi-role auth system. Business owners, customers, and admins each need different views, permissions, and data access. Supabase RLS policies and role-based routing solved this cleanly.

Finally, making the AI tools feel native rather than tacked-on was a design challenge. I integrated them into the existing dashboard sidebar as an "AI Tools" section, using the same design language as the rest of the app.

What Makes It Different

Most "creator platforms" are just storefronts. VendorLite is a complete business operating system:

• Built-in storefront with cart and checkout
• Customer CRM with full order history
• Appointment booking and invoice generation
• Affiliate program with referral tracking
• Blog engine for content marketing
• Real-time analytics dashboard
• AI-powered market research, writing assistance, and web reading
• Two-factor authentication and admin security monitoring

All of this works out of the box. No integrations to break. No data silos. Just log in and start selling.

Built With

• api
• edge-functions)
• gemini-2.5-flash
• intersection
• observer
• postgresql
• react
• shadcn/ui
• storage
• stripe
• supabase-(auth
• tailwind-css
• typescript
• vite