How to use generative AI search with your product catalog

Generative AI (GenAI) is already transforming the way people search. This is huge, given that one way to look at eCommerce at its essence is a search bar. Overall, nearly 60% of searches last year ended without a click. This data suggests that many of these searchers may be getting the information they need from AI. While the same study shows AI overviews for Google haven’t yet had a massive impact on search results, that’s sure to change quickly. As AI answers more of shoppers’ questions up front, their behavior and the structure of their searches will shift everywhere — including on individual retailers’ websites.

It’s impossible to ignore the impact of Amazon’s Rufus and Google Shopping’s latest updates on consumer’s search expectations for their favorite retailers. Virtual shopping assistants now let shoppers ask very detailed questions about products in natural language, and get very detailed responses. With OpenAI’s release of Operator, agentic search is coming to ecommerce, and fast. In the past, things that would have taken multiple steps (like comparing two products or getting clothing recommendations based on the local weather) can happen with a single prompt.

Play Video

Retailers who get ahead of this trend can implement virtual shopping assistants and GenAI search using their own product catalogs and underlying data. We’ve been experimenting with using Elastic Path Product Experience Manager data to create a GenAI product assistant for customers — getting them the personalized shopping experience they’ve come to expect from places like Google and Amazon. Let’s take a look at how it works, and explore how GenAI could change the search dynamics on your website.

Sneak peek: GenAI product assistant within Elastic Path

The experimental GenAI product assistant uses Elastic Path Product Experience Manager as its data source, and is delivered as a feature of the Elastic Path Composable Frontend. Composable Frontend is an open source tool built in Next.js, which means any merchant can use it to either build a full storefront or add on to their current storefront.

To use the agent, upload your catalog to Product Experience Manager and then connect it to Composable Frontend. As long as your products are synced into Product Experience Manager, you can use the agent with your current commerce platform.

From there, the GenAI product assistant uses natural language processing (NLP) to automatically generate a list of questions for every product, based on the product information in Product Experience Manager. Shoppers can either choose from these questions, or create their own prompts to ask other questions they may have about your products. The example below shows how the assistant works for both a pre-generated question and a user-generated question.

Play Video

While this feature is still experimental, we built it to prove that a composable approach is the best way to integrate and test new features within the parameters of your existing commerce system — without extensive custom development work.

The future of commerce search

The bottom line? Search habits are quickly changing. The market share for traditional search engines like Google is threatened by younger generations’ preferences for social media platforms, AI search (e.g. ChatGPT), and voice assistants like Siri. If you take a closer look at these trends, soon the search bar will no longer be the best or even the most prominent way to navigate a commerce site.

Over time, we, as internet users, have learned to conform to search engine rules to get what we want out of the web. Now that finding what we need is as simple as asking a question, these old world search rules and associated behaviors will quickly change.

While the form factor might look different for retailers (e.g. a search bar vs. a conversational assistant), the underlying data remains the same. Your product catalog serves as the best foundation for any successful search experience, including the ones driven by GenAI. Choosing a composable catalog and frontend experience builder means that you can integrate with your existing commerce system via a simple API connection — without having to build these features from scratch.