Inspiration
Millions of small shopkeepers in Pakistan, India, Bangladesh, Africa and Latin America run on store credit ("udhaar"). They track it in paper notebooks that get lost, miscalculated, or ignored, and many feel awkward asking customers for money. Existing ledger apps mainly record debt. We wanted to build one that also helps shopkeepers decide who is safe to give credit to and how to ask for repayment politely.
What it does
CrediLedger AI is a smart credit ledger for micro-merchants.
- Dashboard: total outstanding credit, active debtors, and a recovery rating at a glance.
- Customer ledger: balances for every customer, searchable by name.
- Traffic-light risk status: Green (safe), Yellow (be careful), Red (stop giving credit), so anyone can understand it without reading numbers.
- Voice entry: the shopkeeper taps the mic and speaks, and the spoken text is captured live in the browser.
- Smart reminders: friendly, firm, or formal payment messages based on the customer's risk level, ready to send via WhatsApp.
How we built it
- React + TypeScript for the interface
- Tailwind CSS and Lucide Icons for a clean, accessible design
- Web Speech API for real-time voice recognition, with no API key needed
- Rule-based risk scoring and reminder generation that run in the browser
- Built and deployed with Bolt.new
Challenges we ran into
- Our first voice feature used hardcoded text, so we replaced it with real speech recognition.
- We had no time to set up a backend, so we kept everything client-side for reliability.
- We wanted an interface that a non-technical shopkeeper could use without training.
Accomplishments that we're proud of
A working prototype with live voice input, color-coded risk indicators, and smart reminders, built under a tight deadline.
What we learned
Simple design matters more than complex features when your user is a busy shopkeeper. We also learned how to use browser-native APIs to avoid backend complexity.
What's next for CrediLedger AI
- Real WhatsApp Business API integration for automatic reminders
- Cloud database and login (Supabase) so data is saved across devices
- A trained ML risk model using real payment history
- Multi-language and multi-currency support
- Partnerships with microfinance institutions, so repayment history can help unbanked merchants qualify for small business loans
Business model
Free basic ledger. Premium subscription (about $2/month) for advanced risk insights, automated reminders and reports, plus a small fee on in-app payments.
Built With
- bolt.new
- lucide
- react
- tailwindcss
- typescript
- vite
- web-speech-api
Log in or sign up for Devpost to join the conversation.