BillBuddhi-Local-first AI invoice assistant for Indian small businesses using Gemma, Ollama, Tesseract OCR, and Streamlit.

BillBuddhi is an AI-powered invoice assistant designed specifically for Indian freelancers and small businesses. The tool aims to simplify the tedious task of manually entering invoice details into spreadsheets. BillBuddhi achieves this by automatically extracting invoice information, allowing users to review the results and flag any basic inconsistencies. Once the invoice details have been reviewed and corrected, users can export the reviewed invoices to a CSV ledger.

The project was created as part of the Hacktoberfest Open-Source AI Challenge Week 2, with a focus on helping small businesses save time on manual data entry while maintaining control over their financial records. To build BillBuddhi, the developer used Python and Streamlit as the primary programming languages. Tesseract OCR was employed for extracting text from invoice images, while pypdf was used for extracting text from PDFs.

A local Gemma model, accessible through Ollama, was utilized to suggest structured invoice fields such as supplier name, invoice number, invoice date, and currency.

The final product includes features such as the ability to upload PNG, JPG/JPEG, or WEBP invoice images, or PDFs that contain selectable text. The system automatically extracts text from the uploaded files and uses a local Gemma model via Ollama to extract structured invoice fields. Users are then given the opportunity to review and manually correct the suggested values before adding the invoice to the ledger.

Additionally, the application offers the option to export reviewed invoices to a CSV ledger for further bookkeeping purposes.