'Kal' means yesterday and tomorrow. I built home bakers a WhatsApp order sheet that never guesses which.

Developer Rohit has created 'Pakka,' a privacy-focused web application designed to help home bakers manage WhatsApp orders. The tool uses the Gemma 4 E2B model running locally in the browser via WebGPU to parse chat logs. By focusing on extracting specific data points rather than relying on the LLM for complex date arithmetic, Pakka avoids common pitfalls where models struggle with ambiguous terms like 'kal' (which means both yesterday and tomorrow in Hindi). The application processes data entirely on the user's device, ensuring that sensitive customer information—such as names, addresses, and phone numbers—remains private. By automating the creation of bake sheets and order confirmations, Pakka aims to reduce the manual labor and potential errors associated with managing small-scale food businesses. The project is open-source and demonstrates a practical, privacy-first approach to integrating small language models into everyday workflows.