From WhatsApp to Kitchen: An AI-Powered Order Automation System (mateolafalce.github.io)

Mateo Lafalce published an open-source prototype that automates order-taking for a local pizza shop by combining OpenAI, the WhatsApp API, MySQL, and Mercado Pago. The modular system chains a WhatsApp chat interface (currently runnable only on localhost because real WhatsApp API access requires business validation) to an OpenAI-powered conversational layer that lists products, aggregates line items and subtotals, generates a Mercado Pago payment link, and—after payment—triggers backend-to-frontend notifications so kitchen staff and delivery workflows start and finish the order. The repo exposes endpoints and a simple web frontend (no JWT/user management implemented) so teams can swap domain-specific modules without rewriting core logic. For the AI/ML community this is a practical reference for integrating generative models into real-world transactional pipelines: it highlights prompt-managed product parsing, event-driven order state transitions, and multi-service orchestration (payments, DB persistence in MySQL, messaging). Important caveats include missing production pieces—WhatsApp business verification, authentication, and an OpenAI cost-control mechanism (the author suggests weighting customers to limit token spend). The project is useful as a blueprint for small businesses automating conversational commerce while surfacing implementation concerns around API validation, cost management, and modular extensibility.