Inspiration
Millions of small local shopkeepers (Kirana stores, daily vendors, local retailers) in Tier-2 and Tier-3 towns struggle with complex accounting software and manual paper bahi-khata. Manual bookkeeping leads to lost credit tracking (udhar), calculation errors, and wasted hours. We wanted to build Vyapaar AI to make business management as simple as sending a voice note on WhatsApp.
What it does
Vyapaar AI is a voice-first smart digital ledger and business assistant tailored for local merchants:
- Voice-to-Ledger: Shopkeepers can speak naturally in Hindi or Hinglish (e.g., "Ramesh ko 500 udhar diya"), and the app parses the intent, amount, and customer name into clean ledger entries.
- Automated WhatsApp Reminders: Creates 1-click friendly payment reminder messages embedded with UPI payment links.
- Instant Product Catalog: Allows merchants to snap a photo of any new stock, set a price, and auto-generate clean promotional graphics/catalogs ready for WhatsApp Status.
- Daily Financial Snapshot: Simple, glanceable view of total daily cash flow, receivables, and overall profit.
How we built it
- Frontend / Mobile UI: Designed using Figma and structured for mobile deployment (Flutter / React Native / no-code rapid prototype).
- Voice & NLP Processing: Integrated Speech-to-Text and lightweight AI prompt extraction to parse local conversational Hindi phrases into structured transaction data (Debits/Credits).
- Backend & Database: Supabase/Firebase for authentication, real-time transaction updates, and offline data sync.
Challenges we ran into
- Handling local dialects, colloquial terms, and informal Hindi sentence structures to accurately extract numbers and entity names.
- Keeping the UI ultra-minimal and high-contrast so anyone without technical literacy can operate it with a single tap.
Accomplishments that we're proud of
- Designing an intuitive workflow where a complete ledger transaction takes under 5 seconds with zero typing.
- Bridging high-end AI capabilities to solve a real, everyday grass-roots business problem.
What we learned
- How crucial accessibility and voice-first UX are for non-desk and small retail workers.
- Designing offline-first architectures to handle spotty internet connectivity in rural/semi-urban areas.
What's next for Vyapaar AI
- Adding automated invoice generation and direct ONDC catalog export.
- Support for regional dialects and offline local voice parsing.
