HisabMate: A Local-First, AI-Assisted Digital Bahi-Khata for Indian Microbusinesses
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle What I Built HisabMate is a local-first, AI-assisted bahi-khata (the traditional credit ledger) for Indian microbusinesses. If you have ever bought something from a neighbourhood kirana, a tailor, or a small who
Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle Submission 💸
This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle
HisabMate is a local-first, AI-assisted bahi-khata (the traditional credit ledger) for Indian microbusinesses.
If you have ever bought something from a neighbourhood kirana, a tailor, or a small wholesaler, you have probably seen the system: a notebook where the owner writes who owes what. Customers take goods on credit ( udhaar ), pay back in pieces, and the owner reconciles it all by hand. It works because it is simple, but it is also fragile. Pages get smudged, entries go missing, and a disputed balance turns into an argument between two people who both want to stay on good terms.
Many bookkeeping tools exist, but they often assume cloud accounts, subscriptions, and a comfort with accounting vocabulary that a busy shopkeeper may not have. HisabMate takes a narrower path: keep the ledger on the owner's own machine, make the arithmetic exact and boring, and let a local language model help with the typing, never with the money.
The person who would pay for something like this is the owner who already runs on credit and handwritten pages and wants a clearer, searchable, exportable record without handing customer data to a third party. To be upfront: I have not had a paying user yet, so that is my hypothesis about the market, not a result.
HisabMate is not deployed publicly. It runs locally, and the two screenshots below are from the real application running on localhost.
Screenshot 1: HisabMate running locally at http://127.0.0.1:5174/.
Screenshot 2: a second view of the working local application.
HisabMate is a local-first AI-assisted bahi-khata for Indian microbusinesses The Python backend owns validation, integer-paise calculations, SQLite persistence, idempotency, and CSV export. The React frontend is a local dashboard and review workflow.
No Python package installation is required: the backend uses the standard library. Frontend dependencies are declared in frontend/package.json.
Set-Location.\frontend npm install Set-Location.. Enter fullscreen mode Exit fullscreen mode The backend creates the SQLite schema automatically on first start. It does not seed customers or transactions.
Install Ollama separately, start its local service, and install the configured model:
ollama pull llama3. 2:3b Enter fullscreen mode Exit fullscreen mode HisabMate uses llama3.2:3b by default through http://localhost:11434. Override the model or endpoint when needed:
$ env: OLLAMA_MODEL = " llama3.2:3b " $ env: OLLAMA_HOST = " http://localhost:11434 " Enter fullscreen mode Exit fullscreen mode The…