Social Shopping powered by AI

Make shopping social and fun

Executive Summary

Shopping is a social experience. How often do you go alone to a shopping mall? But, the world of online shopping is very lonely. How can we make online shopping more fun and social?

In this project, we will build a collaborative, intelligent and rich shopping experience.

Phases of Shopping

There are multiple phases in the shopping life cycle. We will try to illustrate the different phases of shopping below.

Shopping Funnel

The shopping life cycle is often triggered by an event or a necessity. For example, my shoe is damaged and I want to buy a new one or my boyfriend’s birthday is coming up; I need to get a gift for him. Once you have identified the reason for the purchase, you look for the latest trends in the market and try to identify what to buy. You then narrowed down your list, you start researching on the products trying to understand the specifications, what the user scenarios etc. On the items you like, you look into user reviews. You will also ask a friend or two their opinion on the product. Once you have narrowed down to the product to buy, you start comparing prices at multiple stores, look at shipping estimates etc. The last step of the process is the actual buy.

As you can see, shopping is a complex process. The user has to use multiple channels to complete a purchase.

We want to explore if AI (artificial intelligence) assisted shopping can make this experience better. The idea is to have a shopping companion who can stay with you through out the process and help you complete relevant tasks.

AI Assisted Shopping Experience

Below, we will illustrate how a conversational bot can make the shopping experience better.

In the below video, two people are trying to buy a camera for an upcoming trip. In the conversation, there is also an intelligent agent who is helping them through the process. First, when they want to get the trends, the agent gets them the latest trends in camera. They research a few cameras from within the conversational window. They are able to get information on the cameras (images, battery life etc) just by using the conversational interface. They look at user reviews and compare prices from different stores and complete the purchase from one of the stores.

Behind the Scenes

In this project, we used Telegram as the conversational interface. The conversational models were trained on wit.ai. We used Amazon Product Search APIs to get the product information. And, we used DataWeave’s Price Intelligence APIs for price comparison.

Contributors

Anup Reghunathan, Shashank Kumar, Deven Bhooshan and Manpreet Singh Bedi

You can follow our work at http://www.botinsight.com