AI Shopping Assistants: OpenAI, Perplexity & Startup Competition

AI-Powered Shopping Features Launched by OpenAI and Perplexity

As the holiday shopping season approaches, both OpenAI and Perplexity have unveiled new AI shopping functionalities this week. These features are integrated directly into their existing chatbot platforms to assist users in researching potential purchases.

Similar Functionality Offered by Both Platforms

The capabilities of these tools are remarkably alike. OpenAI illustrates that users can request assistance from ChatGPT in locating, for example, a “new laptop suitable for gaming under $1000 with a screen exceeding 15 inches.” Alternatively, users can submit images of expensive clothing items and request similar options at a reduced price.

Perplexity Leverages Chatbot Memory for Personalized Recommendations

Perplexity is emphasizing the benefit of its chatbot’s memory in enhancing shopping-related searches. The platform suggests that users can receive recommendations tailored to information the chatbot already possesses about them, such as their location or profession.

Potential Impact on AI Shopping Startups

Adobe forecasts a 520% increase in AI-assisted online shopping this holiday season. This growth could significantly benefit AI shopping startups like Phia, Cherry, and Deft (now known as Onton). However, with OpenAI and Perplexity expanding their presence in AI shopping, are these startups facing increased risk?

The Value of Specialization

Zach Hudson, CEO of interior design shopping tool Onton, believes that AI shopping startups focusing on a specific niche will continue to deliver a superior user experience compared to general-purpose tools like ChatGPT and Perplexity.

Data Quality as a Key Differentiator

“The quality of any model or knowledge graph is directly tied to the quality of its data sources,” Hudson explained to TechCrunch. “Currently, ChatGPT and LLM-based tools such as Perplexity rely on existing search indexes like Bing or Google. This inherently limits their effectiveness to the top results returned by those indexes.” (Perplexity clarified to TechCrunch that it maintains its own search index.)

Search as a Historically Weak Area in Fashion

Daydream CEO and veteran e-commerce executive Julie Bornstein shares this perspective. She noted to TechCrunch earlier this year that search has historically been a neglected aspect of the fashion industry due to its inherent limitations.

The Nuances of Fashion Shopping

“Fashion is uniquely complex and emotionally driven – discovering a dress you adore differs significantly from purchasing a television,” Bornstein stated to TechCrunch on Tuesday. “A deep understanding of fashion shopping requires domain-specific data and merchandising logic that accounts for silhouettes, fabrics, occasions, and how individuals coordinate outfits over time.”

Building Proprietary Datasets

AI shopping startups are investing in the development of their own datasets to train their tools on higher-quality information. This is more readily achievable when focusing on specific categories like fashion or furniture, rather than attempting to encompass all of human knowledge.

Onton’s Approach to Data Cataloging

In Onton’s case, a data pipeline was created to catalog hundreds of thousands of interior design products in a more organized fashion, enabling the training of its internal models with improved data. Hudson contends that startups failing to pursue this level of specialization risk being eclipsed by larger competitors.

The Challenge of Competing with Major Players

“If a startup relies solely on readily available LLMs and a conversational interface, it will be difficult to effectively compete with larger companies,” Hudson asserted.

Advantages Held by OpenAI and Perplexity

OpenAI and Perplexity benefit from an existing user base and the ability to establish partnerships with major retailers from the outset. While Daydream and Phia direct customers to retailer websites for purchases – often earning affiliate commissions – OpenAI and Perplexity have integrated partnerships with Shopify and PayPal, enabling in-app checkout functionality.

Exploring Potential Revenue Models

These companies, which require substantial computational resources, are still seeking sustainable paths to profitability. Drawing inspiration from Google and Amazon, exploring e-commerce as a revenue stream makes sense – retailers could potentially pay for product advertising within search results.

Potential for Exacerbating Existing Search Issues

However, this approach could potentially worsen the existing problems users experience with search.

The Importance of Vertical Models

“Specialized models – whether in fashion, travel, or home goods – will achieve superior performance because they are optimized for genuine consumer decision-making,” Bornstein concluded.

Additional reporting by Ivan Mehta. Updated 11/26/25, 11:30 AM ET with comment from Perplexity.