IBM AI for Travel

For IBM’s Center for Integrated Design course, my team worked with IBM Watson and its AI natural language processors to develop a travel assistant.

Team: Namrata Gupta, Lisa Barson, Vincent Carson, IBM Mentor Claire McCloskey, Annie Xue, and Anisha Oommen
My Roles: design research, content and storytelling

 
IMG_20181204_191116.JPG

————————————————————
Overview

 

Objective

People need a better way to discover, plan and book personalized trips using natural language conversation. Friends and family have opinions, and it’s hard to navigate the enormous number of reviews and social media posts related to destinations. All of this makes it difficult to choose where to go and what to do.

Challenge

How can we integrate Watson technologies to improve a corporate traveler’s experience.

Solution

The Travel Genie is a 20-questions-game experience that helps people with quick, easy, and optimized decisions for meals and events during busy business trips. It utilizes IBM Watson’s natural language processor as well as connected data from other platforms and accounts in order to quickly and effectively help you make the best decision for what to do next.

The user must first answer questions about their hometown preferences, giving the Genie hints as to the user’s personality as well as determine which establishments are most liked by locals when other users come into town.

Then, the user may prompt the Genie with questions like: “Where can I grab coffee on the way to my meeting?”, or the Genie may alert the user when it spots a place nearby, like: “You’re passing a great Polish deli nearby that [another corporate traveler] has reviewed on a previous trip. Would you like to make a stop?”

————————————————————
Process

 

Our Approach

As we were synthesizing pain points from our 12 interviewees, we realized that corporate travelers have an abundance of unique needs and restrictions when it comes to traveling. If we design for this extreme case user, our solutions apply to a broader audience. With corporate travelers, we found that they:

  • plan for diverse groups

  • desire to meet client and boss’s expectations

  • default to known chains or room service

  • want a local experience

  • want to leave work at work

  • often deal with last minute changes

image6.png
 

Personas

Through our research, we created 2 different types of corporate traveling personas along with their need statements.

Howard: 58, Travels more than 75% of his billable hours, works up to 70-80 hours, all work and no play, religiously orders room service.

He needs a way to…

  • find fast, convenient meals so that he doesn’t order in and miss out from a city’s local offerings

  • Howard’s employer needs a way to get Howard to explore the city so that Howard feels greater employee satisfaction and decrease dollars expensed on delivery fees and room service.

Ellie: 26, works as a consultant, travels twice a month for her job, uses travel points for leisure travel, loves to travel the city she’s in.

She needs a way to…

  • get great recommendations fast so that she doesn’t have to wrestle with indecision and can impress her coworkers

  • connect with people in the city so that she feels like she is getting an authentic view of the city and doesn’t feel alone

  • plan impromptu experiences around her work schedule so that she feels like she is maximizing her free time in an unfamiliar, new city.

 

Our Focus

With the synthesized need statements of each persona, we boiled it down to one general statement: corporate travelers need a way to find a unique place to eat in 5 minutes. Our goal was to design a product that is incredibly efficient, usable, and valuable.

This was a tall order, so we decided to simplify this complex function into an approachable experience by framing it as a “20 questions” game. This app would not only give simple, easy, and personalized suggestions, but also almost game-ify the travel planning process, with the fun of a guessing genie and the efficiency and ease of a limited back-and-forth conversation.

image2.jpg
 

Prototyping

To test this out, we got together and played the roles out ourselves. A few of us would pretend to be the “guessing genie” that had to give a good, local suggestion for a place to eat, while the others would pretend to be the business traveler planning a client happy hour after work or grabbing a bagel on the way to work. (i.e., “What are you feeling for dinner today?” “Some Korean food would be nice.”, etc.)

We found a few things:

  • The conversation really only took 3-5 back-and-forth’s, maximum

  • Many of the questions we thought about asking could be done in the AI back-end. For example, instead of our “genie” asking how much time we have to eat or how far we were willing to travel, we could opt to connect to the user’s Google Calendar and email

  • We absolutely needed to increase the usability of a voice assistant, since we were having trouble answering questions in person to our friends, much less a machine.

ibm+3.jpg
Image from iOS-min.jpg

————————————————————
Reflection

 

My Thoughts

Our team went through many rounds of iterations and major pivots. One of the biggest lessons I learned was to not design for myself because there would be an inherent bias that’s not based on research.

Overall, I’m proud of the work our team has done with this app because it has the potential to solve business problems by increasing overall job satisfaction for an employee. Furthermore, we left a lot of room for more features to include, allowing this idea to extend beyond this minimum viable product stage.

MVIMG_20181204_181237.JPG