How AI Search Pushes the Commercial Web to its Edge

Today, online commerce is centered around driving traffic and conversions to websites. Lately, I’ve been thinking, what happens when ChatGPT-style search engines send people answers, rather than links? 

In recent months, both Microsoft and Google released chat-enabled, question-answering search offerings powered by generative AI. Google also recently announced a conversational integration into its Shopping Graph. While some AI chat answers seem cut from the movie “Her,” others are surprisingly useful, compiling information from multiple sources. Frankly, chat answers take some getting used to for experts in “Googling” as we know it. But, it’s not hard to imagine how this change could be useful for high-intent shoppers.

Case in point, a friend of mine was looking for a booster seat for her six-year-old son and spent hours trying to figure out whether he met the height and weight requirements of specific seats. She finally found the answer, embedded deeply within Amazon’s Q&A results, and bought that seat right away. This is a perfect example of a high-intent search that could have been drastically shortened by chat, leading to an immediate purchase.

Microsoft aptly named its ChatGPT-powered browser “Edge.” I believe the commerce world will need to respond to AI with edge commerce. We're already seeing how edge commerce can play out in edge checkout, with social and email “buy now” buttons removing friction from the customer’s point of discovery. Push that a level further, and we may see the commerce website and commerce platform as we know it “unbundle” in response.

AI as an extension of the API economy

Developments such as plugins for ChatGPT fit neatly into a trend that’s been happening behind the scenes in tech — the API economy. While APIs are nothing new, these interfaces are what make modern software components integrate with one another. APIs share information across multiple technologies, many times from different providers.

As venture capital firm A16Z predicted, “traditional, large programs are being broken down into and being offered as — or with data exposed to — services. But more importantly, these pieces and services are being recombined by other companies to create something new… it’s combinatorial innovation in action.”

The problem is most underlying commerce systems are not set up with this connectivity and innovation in mind. Commerce platforms are often walled off and deeply entrenched. They frequently require a middle orchestration layer between APIs and legacy systems operating behind the scenes. These legacy systems might control order management, payments and other business-critical tasks.

This technology workaround manifests itself to the shopper via a broken user experience. For example, a social media user might try to transact on an ad, finding that inventory is sold out on a social platform but available on a website. That’s a lack of integration on full display. 

Other user experience issues abound — from irrelevant product recommendations to “shop the look” pages that don’t redirect shoppers to the exact items and colors in stock. Merchants that want to take advantage of AI chat-based search engines will need to think about the flexibility of their backend commerce systems. Think of the website as a “hub” of information that must be disseminated to a variety of different distribution channels as “spokes.” AI is just one of those channels. APIs must push and pull data efficiently and effectively to each distribution channel.

Get hands on with an Elastic Path Free Trial

Start building the commerce experience your unique business demands with a free Elastic Path Commerce Cloud account. Get in touch with us and we will setup a Free Trial store for you for six weeks.

What will these changes look like for shoppers? 

The primary look and feel of a commerce website might not change for shoppers overnight. Incremental changes tend to happen first.  In fact, some commerce websites have already begun experimenting with ChatGPT integrations to help shoppers find what to buy, or to enhance product descriptions for better conversions.

It’s not hard to imagine that as search habits change among consumers, chat-based search capabilities will be embedded into every commerce website.  I The structure of information behind the scenes will change, so that when a consumer asks a question, it’s not hard for the AI system to retrieve an accurate answer. 

As a result, I imagine that Q&As become more prominent parts of every website. Digital storefronts will be able to replicate the store associate interaction - can I help you today? Data embedded within user reviews might pop up more readily for chat-based searchers. Customer service requests might be more easily resolved online, with robust knowledge-bases feeding into AI-powered customer service answer engines.

These things all exist today but are cumbersome and most often deliver an unsatisfactory experience because they just respond with a link to some other data source that is barely relevant. Suddenly, the AI technology that powers them has become more powerful and accurate, and, most importantly, more relevant to the inquiry, all in real time.

I also expect commerce sites to be able to push and pull their data to various services (powered by social commerce, search, brand partners, marketplaces, and more), so that shoppers can ask questions anywhere and get consistent answers. Shoppers should also be able to check out at the point of discovery, meaning that transactional technologies and underlying catalog infrastructure must be flexible and API-powered, as well. 

AI search is still early in its development, and there are plenty of flaws. Chief among the challenges is misinformation, which makes it hard to trust the results of AI search. Today, it’s still messy. But, much like any technological shift, change happens gradually, then suddenly.

A few decades ago, we couldn’t have imagined that smartphones would work their way so prominently into our daily lives. We’d never bark orders at a speaker to buy paper towels. That’s why it’s not too much of a stretch to imagine shoppers choosing the lowest-friction path toward finding what they need on an increasingly crowded and competitive web. For many, that may mean chatting with machines.