OpenAI has rolled out a new shopping research tool inside ChatGPT just in time for the annual rush of holiday spending, promising faster product discovery, cleaner comparisons, and a more interactive way to sort through deals. While framed as an “expert personal shopper,” the feature largely serves as a structured research assistant: it collects data, asks follow-up questions, filters options, and produces a tidy buyer’s guide. It’s free for all logged-in users across mobile and web, which positions it as a broad, accessible utility rather than a premium upsell.
The tool works much like Deep Research, but tuned specifically for commerce. Users can describe what they’re looking for—anything from a smartphone under a certain budget to a lookalike for a hard-to-find item—and ChatGPT responds with clarifying questions before scanning the web to produce results. It then surfaces sample products during the process, prompting users to mark items as appealing or irrelevant. This swipe-style feedback loop is one of the feature’s most engaging elements, allowing users to quickly steer the research rather than passively wait for an answer.
OpenAI says the system runs on a custom variant of GPT-5-mini trained for shopping tasks, designed to synthesize information from reputable sites, track product criteria, and refine search prompts on the fly. The company is transparent about the model’s limitations, acknowledging that product availability and pricing can still be incorrect and advising users to double-check before purchasing. While the tool aims to outperform the broader ChatGPT models in shopping accuracy, it cannot fully replace the nuance of human judgement—particularly for users who enjoy the ritual of comparison shopping.
In practice, the tool handles a wide range of scenarios: building side-by-side comparisons, finding budget alternatives to premium items, identifying dupes, suggesting gifts based on personal details, or narrowing choices using real-time preference feedback. Its ability to incorporate user context—when personalization toggles are enabled—adds another layer, producing recommendations that reflect prior interactions or stated lifestyle details. For users who prefer speed over endless browsing, this kind of guided research may feel like a practical shortcut.
The process is straightforward. Any shopping-related prompt activates the feature, or users can select it directly from the menu. As the tool collects information, it displays products with quick interest toggles, explains trade-offs, and ultimately assembles a structured guide with links to external retailers. OpenAI says results are generated organically from accessible sites, not through retailer partnerships, and that user chats are not shared with merchants.
Early testers report that the system is intuitive and unusually interactive. It can handle straightforward product requests, such as dog treats for a specific breed, but also more complex gift-giving scenarios that draw on multiple personal clues. Its strengths show up most clearly when a user presents overlapping criteria—budget limits, aesthetics, lifestyle details—where the model can adjust in real time to refine suggestions. While the final recommendations may not always be groundbreaking, they can deliver a practical starting point for purchases, especially for users who value efficiency over spending hours browsing.
Ultimately, the tool isn’t positioned to replace the full shopping process—particularly for those who enjoy the search itself—but it may streamline early discovery and reduce decision fatigue. As OpenAI integrates the feature more closely with its checkout tools, the line between research and purchasing will likely get shorter. For now, it stands as a free, accessible option that adds structure and interactivity to a task many users find overwhelming during the holiday season.
