Garcon

Inspiration

We were at a Chinese restaurant discussing about ideas for Disrupt when we experienced difficulties when ordering food. There were only two menus for the four of us to share, and there were too many dishes for us to choose from. We also found it inconvenient to manually split the bill after dining.

What it does

With Garcon, customers can easily view menus with their mobile phones and spend less time on deciding what to eat. In a restaurant, simply scan the Messenger QR code on your table, choose a "mood" and launch the restaurant menu. Using machine learning, the app intelligently sorts the menu to suit your mood. Garcon also remembers your dietary needs and automatically suggest special food preparation requests. Customers can pay directly for their own food using Venmo.

For restaurants, they can easily use the web platform to upload or change their menu information on the fly and manage incoming orders.

How we built it

The Messenger App is built with Node.js, while the Restaurant Web App is built with Meteor and React. For machine learning, we used a dataset from New York Public Library that records dishes from 1850s to the present. We build a two-layer Long-Short Term Memory (LSTM) Neural Network with word embeddings to accurately predict the mood of a particular dish given the dish name and description.

We also integrated the app with Venmo's API.

What's next for Garcon

Garcon is a fatherly-figure that serves as everyone's personal waiter. We believe that this app can disrupt the dining scene in SF and beyond. Developing a concrete B2B business model and adding a tipping system to Garcon are in the cards. 81% of restaurants in United States still do not use any form of tablet-based ordering system to date (with tablet-menus costing $300-500 per table annually), this represents a multi-billion market opportunity.