How ChatGPT’s New Deep-Search Tool Simplifies Buying
You may be like me and behind on the news, or, actually, I mean buying gifts for family and friends, this holiday season. Well, if you are shopping, Research is a new dedicated experience inside ChatGPT that was designed to build personalized buyer guides. It will be rolling out to all users: Free Plus, Team, and Enterprise, also on the web and mobile app. It is meant to help with complex decisions, like finding the best TV for a bright room or the best gift, rather than just comparing prices. Since it can be used indefinitely, it should be very useful for the holiday season.
When you start a chat, it opens a visual workspace that asks smart, clarifying questions like budget, preferences, and specific needs. The user gets to guide the research by telling the AI model stuff like “More like this” ot “Not interested” to make results better. This feature is powered by a specialized version of GPT-5 Mini trained specifically to read reviews and retail sites. After the model finds out what you like, it will make a summary of top products, with key differences shown, and trade-offs that OpenAI says would normally take hours to research manually. So this feature could be really helpful to many people.
This tool pulls from past conversations; for example, if you are a parent, teacher, or gamer, it would tailor what it recommends. ChatGPT Pulse can proactively suggest accessories based on previous research, and the results are thankfully organic, with no retailer paying to get top results. For now, at least. But even though the AI is more accurate than older models, it can still make mistakes on price and availability, so checking the actual site is recommended. So if you use this feature, be careful with the results! And have a great Holiday season!
Do you think the ability to see the trade-offs of products will make people more confident in their buying decisions? Do you think you will ever use this feature when you are shopping for something? Let us know in the comments!
Source: OpenAI