Biashara Ledger

Biashara Ledger — AI-powered accounting for Kenyan informal traders. Speak your sales, snap receipts, and see your profit instantly. No accounting skills needed.

Comment

Inspiration

What it does## Inspiration

Walking through Kenyan markets, we saw mama mbogas and kiosk owners mentally tracking hundreds of transactions daily with no way to know if they were actually profitable. That gap inspired Biashara Ledger — a tool built for the trader, not the accountant.

What it does

Biashara Ledger is an AI-powered financial tracker for informal traders. Users can:

  • 🎙 Log transactions by voice in Swahili or English
  • 📷 Scan receipts using Gemini Vision AI
  • 📊 View a real-time dashboard showing daily income, expenses, and profit
  • 📵 Work offline and sync when reconnected

How we built it

  • Frontend: Pure HTML, CSS, and JavaScript — no frameworks, no installation
  • AI: Google Gemini 2.0 Flash API for voice parsing and receipt scanning
  • Storage: Browser localStorage for offline-first data persistence
  • Charts: Chart.js for the financial dashboard
  • Deployment: Netlify (one drag-and-drop)

Challenges we faced

  • Parsing mixed Swahili-English voice input accurately using Gemini prompts
  • Building a fully offline-capable app without a backend server
  • Designing a UI simple enough for low-literacy users while keeping it functional

What we learned

  • How to engineer effective Gemini prompts for structured financial data extraction
  • How to use the Web Speech API for multilingual voice recognition
  • That the best solutions are the ones users never have to be taught

What's next

  • M-Pesa transaction sync
  • USSD fallback for feature phones
  • Multi-currency and group savings support

How we built it## How we built it

Built entirely with HTML, CSS, and JavaScript — no backend, no installation required. We integrated the Google Gemini 2.0 Flash API directly from the browser for voice transaction parsing and receipt scanning via Gemini Vision. Chart.js powers the financial dashboard, and browser localStorage handles offline-first data storage. Deployed on Netlify with a single drag-and-drop.

Challenges we ran into

Parsing mixed Swahili-English voice input was the biggest hurdle — we had to carefully engineer Gemini prompts to extract structured financial data from informal, conversational speech. Building a fully offline-capable app without any server was also tricky, as was designing a UI simple enough for low-literacy users without sacrificing functionality.

Accomplishments that we're proud of

We built a fully working AI-powered app in one hackathon night that solves a real problem for millions of Kenyan traders. The voice entry works in Swahili, the receipt scanner extracts items from photos automatically, and the entire app runs in a single HTML file — no setup, no login, no technical knowledge required from the user.

What we learned

We learned how to engineer effective Gemini prompts for structured data extraction, how to use the Web Speech API for multilingual voice recognition, and how to build a production-ready AI app without a backend. Most importantly, we learned that the best solutions are the ones users never have to be taught how to use.

Challenges we ran into

Accomplishments that we're proud of

What we learned

What's next for Biashara Ledger

Built With

Share this project:

Updates