How this shopping chatbot came to life

Recently, there has been a lot of interest in using chatbots as an additional channel to reach consumers where they hang out: in messaging apps. Facebook Messenger, for instance, boasts over 900 million users a month.

Other popular messaging platforms that integrate chatbot technology include Skype, Kik, Slack, and Telegram, just to name a few. Even Google has recently launched Allo, an A.I.-driven smart messaging app.

[aditude-amp id="flyingcarpet" targeting='{"env":"staging","page_type":"article","post_id":2076846,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"bots,commerce,","session":"B"}']

Chatbots are well suited to provide a wide array of services, ranging from basic customer support to specialized information query and retrieval across various business operations. Repetitive tasks are especially well suited — chatbots never tire, and they can provide the same quality of service 24/7. Most importantly, chatbots facilitate an instant, private channel to each end user, so you can enhance the user experience by customizing the interactions based on previously learned behaviors. For example, credentials and preferences can be saved to the database and utilized to predict or recommend next actions tailored to each individual user. Personalization, coupled with a chatbot’s charismatic, nonjudgmental, and forgiving identity, fosters a comfortable intimacy with users.

Chatbots employ a conversational UI, enabling users to easily jump to any point in the task flow, unlike conventional mobile apps, which tend to progress in one direction through a predefined set of screens. By understanding user intent and saving proper context, you can effortlessly jump forward and backward multiple steps at a time, with little overhead. As a mobile app developer, I can certainly appreciate this convenience and the ability to provide intelligent services and relevant content to the user.

Chatbots are platform agnostic and can be accessed from any device with a chat app installed, such as Messenger. This frees the developer from creating and supporting several versions of the same app for multiple devices. Users benefit by having the convenience of accessing services from a single consistent portal independent of device platform, not having to worry about updates.

Design considerations

Before designing a product for a new platform, we need to consider how best to take advantage of the features and capabilities the framework provides. Content sharing among users within messaging apps can be accomplished with very little friction and should be exploited. So, my goal was to build a chatbot that would expose both the social and collaborative features of chat.

5Gifts4Her — a revolution in online gift shopping

The 5Gifts4Her chatbot showcases intelligent gift recommendation, easy product discovery, personalization, social, and collaborative features that make online shopping more enjoyable for an individual or a group.

In order to process natural language queries, I incorporated and trained a Wit.ai instance using both slot-based and flow-based stories to extract intent, predict subsequent actions, and navigate through the conversation.

For the UI design, I decided to employ a hybrid approach and used both text and structured messaging. I wanted to allow new users to quickly become accustomed to the chatbot features via a menu of buttons. Advanced users, on the other hand, could take some shortcuts by typing conversational text requests, like “I need to send some flowers to my girlfriend” to get to flower options or “wish code” to quickly access Wish Code entry. I included quick replies to take the guesswork out of what answer format the chatbot understood best.

Dynamic gift recommendations

Gift recommendations are easily accessible via a menu or by typing out a request. The chatbot will respond by running through a list of queries while navigating a decision tree in the background to instantly build a recipient profile. Next, it will invoke a custom recommendation engine and dynamically query the Amazon server based on that profile and display a list of relevant products in a carousel.

[aditude-amp id="medium1" targeting='{"env":"staging","page_type":"article","post_id":2076846,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"bots,commerce,","session":"B"}']

There’s even a Quick Picks option in the persistent menu for users who need gifts in a hurry, and want to just grab and go.

Dynamic product discovery

There will always be some users who want to browse for something more specific. For this reason, I implemented the Find feature to enable product discovery based on the supplied keywords. For example, “Find pink women’s running shoes” or “Find green lawnmower” would return a carousel of product images and options that fit that description. Brand names can also be used in search queries. Essentially any product on the Amazon online store may be retrieved using the Find feature.

Multiple shareable wishlists

To promote collaborative and social engagement, I provided the ability for each user to save favorite products into one or more wishlists, and invented a unique method to seamlessly share these wishlists with other users of the chatbot via a Wish Code. It only takes two clicks to share a list with someone else. The receiver can then decide to just view the list or save it to their own collection for later viewing. You only need to share the Wish Code once, and any changes made to the list by the owner will immediately be visible to each receiver.

The shareable wishlists facilitate a virtual, social, group shopping experience. It’s almost as if you are shopping with your friends and family at the mall, but instead, you shop at your own location and convenience, exchanging Wish Codes to visualize what the other shopper is looking at.

[aditude-amp id="medium2" targeting='{"env":"staging","page_type":"article","post_id":2076846,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"bots,commerce,","session":"B"}']

Imagine that you need to find a gift for your mom’s birthday and your sisters live in another state. You could still virtually shop together by searching for products individually, adding them to one or more wishlists, and then exchanging Wish Codes to select together the perfect gift for mom. How convenient is that?

These wishlists can also be used to generate and access gift registries for any occasion. Add select items to wishlists and post the Wish Code on invitations and websites for others to retrieve.

In addition, wishlists can be used to store frequently purchased items.

By combining virtual shopping and chat within the same app, you can have a fun, social experience with friends without having to leave your home. The UI is minimalist, clean, and uncluttered compared to traditional online shopping apps — there are no ads or other distracting images displayed on the screen. This allows users to focus on their tasks and accomplish their shopping goals more effectively.

[aditude-amp id="medium3" targeting='{"env":"staging","page_type":"article","post_id":2076846,"post_type":"guest","post_chan":"none","tags":null,"ai":false,"category":"none","all_categories":"bots,commerce,","session":"B"}']

New trend?

Will this be the shopping trend of the future? Will online shopping soon be dominated by A.I.-powered intelligent chatbots or virtual agents? Perhaps one day we could completely trust the chatbot to select, purchase, and ship a gift directly to a recipient without additional intervention from the user. As chatbots become increasingly intelligent, driven by improved natural language understanding and deep learning techniques, users may become more accustomed to the idea of shopping through a chatbot, and this concept may become a reality in the near future.

You can access the 5Gifts4Her chatbot at m.me/5Gifts4Her.