Sangu - WhatsApp Finance Tracker

Overview

Sangu addresses app fatigue by bringing personal finance tracking to WhatsApp—an app users already use daily.

No need to install another app. No command memorization required. Just chat naturally about your expenses.

Key Features

LLM-Powered Natural Language Processing

• Uses Gemini AI with LiteLLM for conversational expense tracking
• Tool calling enables natural language input processing
• Automated expense categorization

Core Functionality

• Budget tracking and monitoring
• Analytics dashboard for spending insights
• Midtrans payment integration
• Real-time expense logging via WhatsApp

Production-Grade Infrastructure

• Deployed on k3s cluster
• Full CI/CD pipeline with Drone
• Complete observability stack:
  • OpenTelemetry for distributed tracing
  • Loki for log aggregation
  • Grafana for monitoring and visualization

Tech Stack

• Backend: Golang (Gin framework)
• Frontend: React with TanStack Query
• Database: PostgreSQL
• AI: LiteLLM, Gemini Model
• Infrastructure: k3s, Drone CI
• Payment: Midtrans
• Observability: OpenTelemetry, Loki, Grafana

Why This Project?

Traditional finance tracking apps suffer from:

• App fatigue (users don’t want another app)
• Complex interfaces
• Manual data entry

Sangu solves this by:

• Meeting users where they already are (WhatsApp)
• Using natural language (just chat normally)
• Automating categorization with AI

It’s a production-ready demonstration of how LLMs can make everyday tasks more accessible and user-friendly.

P.S. Check out the live project at sangu.mahendrabp.id