The race to integrate artificial intelligence into the online shopping experience is heating up, with OpenAI and Perplexity launching new AI shopping features within their chatbots this week. These tools aim to assist consumers with product research and purchasing decisions, potentially reshaping how people discover and buy goods online, especially as the holiday shopping season begins.
Both companies are offering similar capabilities: users can describe desired products using natural language – for example, “a new laptop suitable for gaming under $1000” – or even upload images to find comparable items at different price points. This move places them in direct competition with a growing number of startups focused on AI-powered commerce.
The Rise of AI Shopping and the Threat to Niche Startups
Adobe recently predicted a 520% increase in AI-assisted online shopping this holiday season, signaling a massive potential market. This growth could significantly benefit specialized AI shopping startups like Phia, Cherry, and Onton (formerly Deft). However, the entry of tech giants like OpenAI and Perplexity raises questions about the long-term viability of these smaller players.
Zach Hudson, CEO of interior design platform Onton, believes that specialization will be key to survival. He argues that general-purpose AI tools rely heavily on existing search engine indexes, limiting their effectiveness. “Any model or knowledge graph is only as good as its data sources,” Hudson explained. “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.” Perplexity clarified to TechCrunch that it does maintain its own search index.
The Importance of Domain-Specific Data
Julie Bornstein, CEO of Daydream and a veteran of the e-commerce industry, echoes this sentiment. She points out that traditional search has historically struggled with the nuances of certain product categories, particularly fashion. “Fashion… is uniquely nuanced and emotional — finding a dress you love is not the same as finding a television,” Bornstein stated. She emphasizes the need for “domain-specific data and merchandising logic” to truly understand consumer preferences in areas like apparel.
AI shopping startups are investing in building their own proprietary datasets, allowing them to train their models on higher-quality, more relevant information. This is easier to achieve in focused areas like fashion or furniture than attempting to encompass all of human knowledge. Onton, for example, has developed a data pipeline to catalog interior design products with greater accuracy.
However, Hudson cautions that startups relying solely on off-the-shelf large language models (LLMs) and conversational interfaces may struggle to compete with larger companies possessing greater resources. The competitive landscape in e-commerce is already fierce.
Advantages for OpenAI and Perplexity
OpenAI and Perplexity have a significant advantage: an existing user base already familiar with their platforms. Additionally, their scale allows them to forge partnerships with major retailers more easily. While companies like Daydream and Phia typically redirect customers to retailer websites for purchases, earning affiliate revenue, OpenAI has partnered with Shopify and Perplexity with PayPal, enabling in-app checkout functionality.
These companies are still navigating the path to profitability, a challenge compounded by the high costs of operating large AI models. One potential revenue stream, mirroring strategies employed by Google and Amazon, is to allow retailers to pay for prominent placement within search results. However, this could potentially replicate the issues consumers already experience with traditional search advertising.
Bornstein believes that vertical, specialized models will ultimately outperform general-purpose solutions. “Vertical models — whether in fashion, travel, or home goods — will outperform because they’re tuned to real consumer decision-making,” she said.
The future of online retail will likely involve a blend of these approaches. The next year will be crucial in determining whether specialized AI shopping startups can maintain their niche or if larger companies will dominate the market. Monitoring the development of proprietary datasets, the success of in-app checkout features, and the evolution of advertising models within these platforms will be key indicators of the direction the industry takes.
