Aria Flights (Concierge Intelligence)

GIF

Inspiration

Having just spent the better part of the last few weekends travelling across North America (Boston, Toronto, Kingston, New Haven, New York, etc.) Travelling can be a pain. Slow check-ins, reminders, packing are annoying enough - not to mention stay on airline websites for hours to find a reasonable price. We decided to use JetBlue's api to put some transparency into the system - one that we think will benefit both parties.

What it does

Aria is an intelligent travel agent that uses your already available data (social media, past purchases, friends, etc.) to determine and recommend flight packages to the customer. Based on keywords (say you want to find a romantic destination... but also on a budget of $600), we slowly filter the list of available flights through a seamless and natural human conversation.

Additionally, we have the ability to plan trips between people, and bring them in on the conversation with Aria, regardless of their departing destination. Being in a long distance relationship can be hard, and current trip planners never account from multi-destination points. We used Facebook to easily leverage your personal and social data to fly easily with your friends (or girlfriend/boyfriend).

One of the biggest challenges to airlines today is filling the remaining 15% of airplanes. Unfortunately standby tickets are largely unavailable to the general public due to more restrictive rules at customs, leading to longer waiting times. We created the Aria's purchasing system to allow two kinds of ticket purchasers - a commited buyer, and an "aria flyer". Commited buyers are able to purchase and use the tickets on the day of, while Aria flyers choose to join group flights that they only have to commit to if it reaches the group minimum. It's a great way to encourage spontaneous travelling and customer engagement.

How I built it

We used the MEAN stack to shell out most of our application. Our team split in half to develop front-end and back-end, because each was vital to creating a "human experience".

Front-end used Sketch, Photoshop, Skeleton, HTML/CSS/Javascript + JQuery and Google Chrome DevTools (<3) to hash out most of our prototyping.

Back-end used MongoDB, Node, Express, and the Jetblue APIs alongside experimental Machine Learning functions to help us understand Natural Language Processing.