Voice-Commerce
A bilingual, voice-first agricultural marketplace connecting farmers, buyers, and transporters directly through conversational AI — crop recognition, voice-guided commerce, and real-time logistics in one ecosystem.
shipped as one MVP
voice + interface support
timeline, farmer to doorstep
Digital Literacy Was the Real Barrier, Not Access
Rural farmers in Andhra Pradesh face significant profit loss due to reliance on intermediaries, driven primarily by low digital literacy. Buyers lack direct access to fresh produce with verified provenance, while transporters have no structured way to discover delivery opportunities. The challenge was to build a bilingual (Telugu/English) agricultural marketplace that eliminates these barriers through conversational AI and voice-first design — making digital commerce accessible to users with minimal typing skills.
- Profit loss — farmers lose margin to intermediaries with no direct buyer access
- Low digital literacy — typing-first interfaces exclude the exact users who need them most
- No provenance — buyers can't verify where fresh produce actually comes from
- Unstructured logistics — transporters have no organized way to discover delivery jobs
"The barrier was never internet access — it was the keyboard. Voice-first design was the only way in."
Four-App Ecosystem on a Unified Backend
A Full MVP, Shipped and Tested in the Field
Full MVP shipped — four production-ready applications deployed and tested on Android devices (Samsung Galaxy M35 5G, Android 14; Samsung Galaxy Note 9, Android 10). Farmers get voice-guided listing creation via camera-based crop recognition plus market price suggestions, with a manual fallback form and expiry-controlled listing management.
Buyers get voice search, a visual catalog with farmer storytelling, cart grouped by farmer, checkout with phone OTP auth deferred until purchase, and order tracking across a 5-step real-time timeline. Transporters get a job board with haversine distance and pay estimates, one-tap acceptance, a 4-step delivery workflow, an earnings dashboard, and turn-by-turn OSRM navigation.
- Voice-first interaction model eliminates digital literacy barriers entirely
- Users can list, search, and transact through conversational AI in their native language
- Admin dashboard gives complete operational control with knowledge base ingestion for AI Q&A
- All four apps share one Supabase backend — no data duplication across roles
Lead Developer / Architect
- Designed and implemented the full-stack system from concept to deployment
- Owned the frontend architecture across four React/Vite TypeScript applications
- Designed the Supabase backend schema and migrations
- Built Edge Functions — Gemini token minting, server-side distance search-listings, delivery job creation, notification dispatch
- Built the voice integration layer — Gemini Live WebSocket client + PCM audio processing via the Web Audio API
- Built the design system (4000+ lines of earth-tone CSS) and implemented full bilingual (Telugu/English) support