With holiday shopping right around the corner, both OpenAI and Perplexity this week announced AI shopping capabilities that integrate with existing chatbots to help users research potential purchases.
These tools are very similar to each other. OpenAI suggests that users can ask ChatGPT for help finding “a new gaming-friendly laptop with a 15-inch screen or larger for under $1,000,” or they can share photos of high-end clothing and ask for something similar at a lower price point.
Meanwhile, Perplexity is researching how chatbots’ memories can enhance users’ shopping-related searches, suggesting that chatbots could ask for recommendations tailored to information they already know about users, such as where they live or what their job is.
Adobe predicted that AI-assisted online shopping will grow 520% this holiday season. This could be a boon for AI shopping startups like Phia, Cherry, and Deft (rebranded to Onton). But as OpenAI and Perplexity push the AI shopping experience further, are these startups at risk?
Zach Hudson, CEO of interior design shopping tool Onton, believes that AI shopping startups with specialized niches will provide users with a better experience than general-purpose tools like ChatGPT or Perplexity.
“The model or the knowledge graph is determined by its data source,” Hudson told TechCrunch. “Currently, LLM-based tools like ChatGPT and Perplexity piggyback on existing search indexes like Bing and Google, so they can only really give you results that are as good as the first few results returned by those indexes.”
Daydream CEO and longtime e-commerce executive Julie Bornstein agrees. She told TechCrunch over the summer that she always thought of search as the “forgotten child” of the fashion industry because it never really worked.
tech crunch event
san francisco
|
October 13-15, 2026
“Fashion… has its own nuances and emotions. Finding your favorite dress is not the same as finding a TV,” Bornstein told TechCrunch on Tuesday. “This level of understanding of fashion shopping comes from domain-specific data and merchandising logic that captures silhouettes, fabrics, occasions, and how people construct their outfits over time.”
AI shopping startups are developing their own datasets to ensure their tools are trained on high-quality data. This is easier to achieve when trying to create a catalog of fashion or furniture, rather than a collection of all human knowledge.
For Hudson, Onton developed a data pipeline to catalog hundreds of thousands of interior design products in a cleaner way and helped train internal models with better data. But if AI shopping startups don’t pursue that level of expertise, Hudson believes they will inevitably be overshadowed.
“It’s very difficult to see how startups can compete with large companies if they are only using off-the-shelf LLMs and conversational interfaces,” Hudson says.
But the advantage of OpenAI and Perplexity is that customers are already using their tools. Additionally, its large presence allows it to sign deals with major retailers from the get-go. Daydream and Phia redirect customers to the retailer’s website to complete the purchase and potentially earn affiliate revenue, while OpenAI and Perplexity partner with Shopify and PayPal, respectively, to allow users to checkout within a conversational interface.
These companies rely on vast amounts of expensive computing power to run their businesses and are still finding their way to profitability. If they are taking inspiration from Google and Amazon, it makes sense to look to e-commerce as an option. Retailers would be able to pay them to promote their products within search results.
But in the end, it may only exacerbate existing problems that customers have with search.
“Models specific to sectors such as fashion, travel, and home goods will perform better because they are tailored to real consumer decisions,” Bornstein said.
Additional reporting by Ivan Mehta.
