With holiday shopping approaching, both OpenAI and Perplexity announced new AI shopping features this week. These tools are integrated into their existing chatbots to help users research potential purchases.
The new tools from both companies are remarkably similar. OpenAI suggests users could ask ChatGPT to find a new gaming laptop under one thousand dollars with a screen over fifteen inches. Alternatively, a user could share a photo of a high-end clothing item and request a similar product at a more affordable price.
Perplexity is emphasizing how its chatbot’s memory can enhance shopping searches. The tool can provide recommendations tailored to information it already knows about the user, such as their location or profession.
Adobe has predicted that AI-assisted online shopping will grow by five hundred and twenty percent this holiday season. This growth could benefit AI shopping startups like Phia, Cherry, or Deft. However, with major players like OpenAI and Perplexity advancing their own shopping experiences, a question arises about the future of these smaller startups.
Zach Hudson, the CEO of the interior design shopping tool Onton, believes that specialized AI shopping startups will continue to offer a superior experience compared to general-purpose tools. He stated that any model is only as good as its data sources. He explained that current tools like ChatGPT and Perplexity rely on existing search indexes, which limits them to the quality of the first few results from those indexes.
Daydream CEO and e-commerce executive Julie Bornstein agrees with this perspective. She previously remarked that search has always been the forgotten child of the fashion industry because it never worked very well. She added that fashion is uniquely nuanced and emotional. Finding a dress you love is not the same as finding a television. That level of understanding for fashion shopping requires domain-specific data and merchandising logic that grasps silhouettes, fabrics, occasions, and how people build outfits over time.
AI shopping startups often develop their own specialized datasets. By focusing on a specific niche like fashion or furniture, they can train their tools on higher-quality data rather than attempting to catalog the sum of all human knowledge. For example, Onton built a data pipeline to catalog hundreds of thousands of interior design products in a clean manner, which helps train its internal models with better information.
Hudson thinks that if startups do not pursue this level of specialization, they risk being overshadowed. He said that if a startup is only using off-the-shelf language models and a conversational interface, it is very hard to see how it can compete with larger companies.
The advantage for OpenAI and Perplexity is their existing user base. Their large presence also allows them to form partnerships with major retailers from the beginning. While startups like Daydream and Phia often redirect customers to retailer websites, sometimes earning affiliate revenue, OpenAI and Perplexity have integrated checkout systems through partnerships with Shopify and PayPal, allowing purchases within the chat interface.
These large AI companies, which require massive amounts of expensive computing power, are still seeking a path to profitability. Taking inspiration from Google and Amazon, e-commerce presents a logical opportunity. Retailers could pay to advertise their products within the AI search results.
However, this approach could eventually worsen the existing problems customers experience with online search. Bornstein believes that vertical models tailored to specific industries like fashion, travel, or home goods will outperform because they are tuned to real consumer decision-making.
Additional reporting by Ivan Mehta.