VentureBeat presents: AI Unleashed - An exclusive executive event for enterprise data leaders. Network and learn with industry peers. Learn More

Can an AI really adapt to your preferences? That’s one of the big questions for the rest of this year and into 2018. Personalization in general is difficult because we’re all different. For shopping to become more personal, it’s even harder because we’re all different shapes and sizes, and we prefer specific brands and experiences.

That’s what makes the eBay ShopBot so interesting. At MB 2017, the company explained some of the machine learning characteristics of this well-known bot, which debuted in October 2016 on Facebook Messenger. Users can browse curated suggestions, but it almost seems like there is a human agent sending you the suggestions. It’s a good example of how shopping general will change in the next few years, especially as ecommerce keeps expanding.

“The new norm for ecommerce will be situational,” explains Japjit Tulsi, the VP of Engineering at eBay. “Some shopping occasions will take place via voice interfaces, without the need for a screen. Some will be a hybrid experience, which could include chatbots. Some will be in virtual and augmented reality. And some will happen through traditional self-serve online experiences without the need for assistance or conversation.”

“Our goal is to build an AI platform that enables truly personalized shopping and provides a service that is better than what the best human sales assistants can offer customers today,” he added.


AI Unleashed

An exclusive invite-only evening of insights and networking, designed for senior enterprise executives overseeing data stacks and strategies.


Learn More

Two good examples of this have to do with how the AI in the ShopBot “remembers” you. It can keep track of your shirt size or the brands you like, so it won’t keep suggesting Nike when you prefer Adidas. The bot also uses computer vision — it can find similar products it knows you like based on a similar image (or an exact photo match).

For now, the bot is in beta but available to any user.

“We’re finding some interesting things,” says Tulsi. “Search engines have made us think before we put something in a box. Users give us short-termed queries, but we want them to give us longer queries.”

The good news: Users don’t need to always describe what they want; they can take a photo and have the bot find similar items. The shopping cart is also AI-powered. Tulsi says the bot can help users track a shipment and receive notifications.

Another feature: ShopBot’s “hotness” indicator has nothing to do with Tinder, the dating app. EBay’s bot knows what has become a bestseller and can feed recommendations that are not only popular but can ship out quickly, are priced low enough, and are highly rated — eliminating endless product searches.

The good news for eBay fans? The bot can scan through over 1 billion products, which might be one of the best examples of how AI will help us reduce all of the complexity in shopping and dealing with any massive data set — especially if the goal is to find a new pair of shoes.