Why I Built BolKharcha

Most finance apps assume you want to fill forms. In Nepal, people talk about money the way they talk about life: “salary aayo,” “esewa ma pathaye,” “phone kine.” BolKharcha started from that gap between how software expects input and how humans actually speak.

I did not want another clone of a Western budgeting template with a Nepali flag sticker on it. I wanted a product that understood local wallets, mixed languages, and the emotional friction of tracking every small expense. That product became BolKharcha — an AI-powered personal finance manager with chat at the center.

The first version was ugly and incomplete, but it could already turn a sentence into a draft transaction. That single loop — speak or type, confirm, save — proved more valuable than a polished dashboard with empty data. Users care about speed to honesty: how fast does the app reflect real life?

Building locally also meant designing for Cash, Bank, eSewa, Khalti, and IME Pay as first-class accounts. Transfers between wallets are everyday behavior here. If your data model only knows “checking” and “savings,” you are already lying to the user.

Under the hood, React Native and Expo let me ship cross-platform. TypeScript kept mobile and API contracts aligned. Node.js and PostgreSQL held the truth of balances. Ollama powered conversational understanding without forcing every request through a black-box cloud API.

The biggest lesson was not technical. It was product taste: local context beats generic fintech every time. When software speaks your language — literally and culturally — money tracking stops feeling like punishment and starts feeling like clarity.

BolKharcha is still evolving. Features like budgets, reports, party loans, and multi-auth are not decorations; they are answers to real questions Nepali users ask. Ship the conversation first. Earn the right to show charts later.