With the holiday shopping season on the horizon, AI giants OpenAI and Perplexity have both unveiled new AI shopping assistant features this week. Integrated into their existing chatbots, these tools aim to help users research potential purchases. However, founders of specialized AI shopping startups remain unfazed, arguing that general-purpose models are too broad to deliver truly personalized experiences, despite Adobe predicting a 520% growth in AI-assisted online shopping this holiday season.

The new AI shopping tools from OpenAI and Perplexity share striking similarities. OpenAI suggests users can ask ChatGPT for assistance in finding specific items, such as "a new gaming laptop under $1000 with a screen over 15 inches," or upload photos of high-end clothing to request similar, more affordable options. Perplexity, on the other hand, highlights its chatbot’s memory capabilities to enhance shopping searches. This allows for personalized recommendations based on existing user data, like location or profession.

Specialized Startups Confident in Niche Approach

Despite the entry of tech behemoths, specialized AI shopping startups like Phia, Cherry, and Deft believe their niche focus offers a superior user experience. Zach Hudson, CEO of interior design shopping tool Onton, emphasized this point to TechCrunch, stating,

Any model or knowledge graph is only as good as its data sources. Right now, ChatGPT and LLM-based tools like Perplexity piggyback off existing search indexes like Bing or Google. That makes them really only as good as the first few results that come back from those indexes.

Julie Bornstein, CEO of Daydream and a veteran e-commerce executive, echoed this sentiment. She previously remarked to TechCrunch that search has long been the "forgotten child" of the fashion industry due to its poor performance. Bornstein further explained,

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 comes from domain-specific data and merchandising logic that grasps silhouettes, fabrics, occasions, and how people build outfits over time.

Specialized AI shopping startups differentiate themselves by developing proprietary datasets, training their tools on higher-quality, domain-specific information. This approach is more feasible when cataloging specific categories like fashion or furniture, rather than attempting to encompass all human knowledge. For instance, Onton has built a data pipeline to meticulously catalog hundreds of thousands of interior design products, providing superior data for its internal models. Hudson warns that startups failing to pursue such specialization will likely be overshadowed.

If you’re using only off-the-shelf LLMs and a conversational interface, it’s very hard to see how a startup can compete with the larger companies,

he added.

OpenAI and Perplexity's Strategic Advantages

However, OpenAI and Perplexity hold significant advantages. Their vast customer bases already use their core tools, and their market presence allows them to secure partnerships with major retailers from the outset. While many specialized startups, such as Daydream and Phia, redirect customers to retailer websites to finalize purchases (sometimes earning affiliate revenue), OpenAI and Perplexity have established direct partnerships with Shopify and PayPal, respectively. This enables users to complete transactions directly within the conversational interface, streamlining the shopping experience.

These AI giants, which rely on immense and costly computing power, are actively seeking paths to profitability. Drawing inspiration from Google and Amazon, e-commerce presents a viable option, where retailers could pay to advertise their products within search results. However, this approach could also exacerbate existing customer frustrations with online search, potentially prioritizing paid placements over genuinely relevant results.

Ultimately, the debate centers on whether general-purpose AI can truly match the depth of specialized solutions. Bornstein firmly believes,

Vertical models — whether in fashion, travel, or home goods — will outperform because they’re tuned to real consumer decision-making.

The future of AI shopping may well depend on which approach can best balance broad accessibility with nuanced, personalized understanding.

Additional reporting by Ivan Mehta.