You are a busy person. You’re at work and you’re hungry. But your fingers are in too deep on that latest TPS report to take a break. You need a taco, and you need it now. Enter the TacoBot, the first-ever food ordering tool integrated with corporate communications tool, Slack. That’s right, now you can discuss this month’s sales figures and order a Quesalupa combo at the same time.
Slack users will be able to order menu items from their local Taco Bell without ever leaving the messaging platform. The TacoBot, created by agency Deutsch, uses artificial intelligence for group and single ordering through natural language. Taco Bell has long been on the edge of fast food innovation, particularly with its latest mobile ordering platform, and the company’s director of digital innovation and on demand Lawrence Kim says the goal is to always evolve its experience and offerings based on the changing consumers consumer behavior. “The TacoBot adds another touch point for fans to access our brand on their terms–when and where they want it most,” says Kim. “Whether it’s through our mobile app, ta.co or now through Slack, we’re always striving to provide access and convenience every step of the way.”
Deutsch’s senior vice-president and creative technology director Martin Legowiecki says the idea came about while they were building Taco Bell’s new site ta.co, and realized the e-commerce capability could efficiently be extended beyond a website or a mobile app. “We also started using Slack as our main communication tool, fell in love with it, and fairly quickly wished we could just order tacos directly through Slack,” says Legowiecki. “The project was born.”
But, as any taco-loving human can imagine, building a TacoBot isn’t easy. Legowiecki says the three most significant challenges were how to translate the visual interface consumers are used to on ta.co to a text-only environment of Slack, how to understand the different ways people can say something that are all the same intent, and how to create a platform that can not only scale to a massive amount of teams and users already on Slack, but also be flexible enough to support other channels.
For Legowiecki, the best way to translate the Taco Bell interface to text was with natural language so people could talk to the TacoBot in Slack just as they would to Janice in accounting or Steve in sales. “We used Wit.ai natural language processing power to solve that,” says Legowiecki. “It’s also built to be part of a larger platform with the potential to connect Taco Bell to other channels like FB Messenger, Amazon Echo, or Apple TV.”
While now only working in Slack, the TacoBot is currently in private beta with several companies, including SAV Studios, Thought Catalog, Giphy, Fullscreen, and FoodBeast, with a wider rollout expected in the coming months.