We're big fans of giving the open space and time to breathe life into creative innovation. There's a lot of focus on the day to day, but it's important to step back and think even more future-forward.

The YML Hackathon is one of our favorite ways to do just that. Every year we encourage employees around the world to break away from client work and dedicate 24 hours to creating with cross-functional teams.

Last year, our teams designed Kontrol, a new app for Tesla. This year? The bartender of the future. 

Niq is an intelligent bartender who lives inside of a motorized bartending station. He can recognize people and greet them by name, tell corny bartender jokes, give drink recommendations, and take orders for your favorite cocktails without a click of a button. Oh, and did we mention he has a British accent?

“Tapping buttons is just so old school,” says Sr. Product Manager Steven McMurray. Steven worked alongside a team of 2 engineers, 2 designers, our head of recruiting to build and design Niq. “ We built this as a proof of concept to show how the marriage of technologies such as Machine Learning, Computer Vision, Artificial Intelligence, IoT etc. will come together to create amazing user experiences in the near future.” 

How it works

When you walk up to the device, Microsoft vision facial verification API’s allow Niq to recognize faces and greet people by name. Siri translates speech to text to help Niq process what the user is saying. APi.ai was used to create the interaction model that takes the text from Siri and makes sense of it, allowing Niq to respond appropriately.

To make Niq feel more human, AWS Polly helped to turn text into lifelike, customizable speech using deep learning. Once Niq gets a command back from Api.ai that a user wants a drink, an Arduino board takes commands via bluetooth from the iPad to automatically start pouring the proper combination of alcohol and mixers into your glass. See the prototype in action:

[vc_video link="https://vimeo.com/226803603" align="center"]

Each time users order a drink, Niq saves their preferences and can be programmed to cut people off when they've had too many. While Niq is just a 24-hour prototype as of now, Niq 2.0 (see mock-ups below) will use machine learning to generate a recommendation engine that can suggest drinks that people with similar tastes might enjoy.