Agentic commerce for independent African sellers — describe your business, Gemini builds the whole store, buyers shop by conversation.
# MarketMind — Agentic Commerce Platform
> Tell AI what you sell. It builds your full store. Buyers shop by conversation, not by scrolling.
MarketMind is a full-stack agentic commerce platform built for independent African sellers. Sellers describe their business in plain language and Gemini AI generates their entire store — product listings, SEO copy, descriptions, and pricing. Buyers shop through a conversational AI agent instead of a category grid.
Built for the **Hack0 Hackathon** · AWS Aurora PostgreSQL + Vercel + Google Gemini
---
## System Architecture
Five layers: **Client** (buyer + seller browsers) → **Next.js 16 App** (SSR + RSC streaming on Vercel Edge) → **API Routes** (10 route handlers for AI, products, stores, orders, auth) → **AI Layer** (Gemini 2.5 Flash for chat + structured output, Imagen for product photos, text-embedding-004 for semantic search) → **Infrastructure** (Aurora PostgreSQL with IAM auth + Vercel Blob CDN). The cross-vendor buyer agent is the centrepiece: a single streamed Gemini response with three tools (`search_products`, `compare_products`, `get_product_detail`) querying every seller's catalogue simultaneously.
---
## Screenshots
### Landing Page
> "Your shop, built by AI." — The hero makes the value prop immediately clear. A marquee ticker runs platform benefits below the fold.
---
### Marketplace
> A category-filtered product grid with a global search bar and an "Ask Agent" shortcut that drops straight into the AI chat. Sellers from across the platform appear here in one unified catalogue.
---
### AI Shopping Agent
> Buyers describe what they want in plain language. The agent searches, filters, and compares products from every seller on the platform — no scrolling, no category hunting. Powered by Gemini 2.5 Flash with `search_products`, `compare_products`, and `get_product_detail` tools.
---
### Seller Dashboard
> The seller overview shows total revenue, store views, active product count, and open orders …