OpenAI представила инструмент Shopping Research — персональный «экспертный шопер», который помогает искать товары, сравнивать модели и находить предложения.

🤖 Как работает
📌 Использует GPT-5-mini, обученный специально под товарные запросы
📌 Читает проверенные сайты и синтезирует данные
📌 Уточняет параметры через диалог
📌 Показывает примеры товаров в режиме «лайк/дизлайк»
📌 Формирует персональный шопинг-гайд с ссылками

📱 Где доступно
Функция работает у всех залогиненных пользователей ChatGPT — Free, Go, Plus, Pro — в мобильных приложениях и на вебе.

🔍 Что умеет
• Подбирать товары по заданным критериям
• Ищать аналоги и «дюпы»
• Строить списки сравнений
• Находить актуальные скидки
• Подбирать подарки
• Работать с фото (например, искать одежду по изображению)

⚙️ Особенности модели
🟣 Обучена избегать спам-сайтов
🟣 Делает перекрёстный поиск по многим источникам
🟣 Может ошибаться в цене и наличии — OpenAI просит перепроверять
🟣 Не передаёт историю переписок ритейлерам

🧪 Практический опыт тестирования
• Инструмент ведёт диалог, задаёт вопросы, уточняет бюджет
• Предлагает товары в формате свайпов — быстрый отбор
• Формирует подробный персональный гид
• Хорошо подходит для подарков, одежды, товаров для животных
• Можно загружать фото — работает как «смарт-поиск по картинке»

🎁 Сильные стороны сценариев
👌 Поиск вещей по фото (как в примере с свитером из Succession)
👌 Быстрый подбор под бюджет
👌 Список вариантов с объяснением плюсов и минусов
👌 Выявление ценовых альтернатив
👌 Удобный инструмент для подарков с несколькими вводными

📉 Ограничения
• Не заменит самостоятельный глубокий поиск — особенно если важна нюансная экспертиза
• Может предлагать слишком очевидные варианты при большом количестве вводных
• Иногда даёт избыточные тематические подборки (например, много Mickey-товаров)

Подробности от ZDNET

ChatGPT’s new shopping research tool is fast, fun, and free — but can it out-shop me?

Follow ZDNET: Add us as a preferred source on Google.

ZDNET’s key takeaways

• OpenAI released a shopping research feature in ChatGPT.
• The experience helps you find products and deals.
• It is rolling out on mobile and web for all logged-in ChatGPT users.

The holiday season is upon us, and the biggest shopping event of the season, Black Friday, is just a few days away. In response, OpenAI launched a new feature in ChatGPT designed to help you with your shopping tasks — for free.

While the company has previously launched AI agents focused on automating the order-placing process, such as OpenAI Atlas and Instant Checkout, shopping research focuses on helping you find what you want to purchase in the first place, acting as an «expert personal shopper.»

Also: Best Black Friday deals 2025: 55+ deals on TVs, laptops, streaming, and more

Similar to Deep Research, when prompted with a product description, ChatGPT will now sift through the internet to put together a guide for you. It will also ask you a series of clarifying questions, using the context from past conversations, and considering product reviews to develop your guide.

What can shopping research do (and how it works)

Shopping research is designed to act as an assistant that can create a personalized shopping experience tailored to your specific criteria and needs in just a few minutes. OpenAI said research outputs can help with a variety of different tasks, including finding a product that meets specific criteria, for example, «help me find a smartphone with 18+hours of battery life under $1,500.»

Other examples include finding dupes or lookalikes of a product, comparing different products with a detailed trade-off list that is catered toward your specific needs, finding product deals, and helping you choose gifts for people on your list.

Also: Why Amazon really doesn’t want Perplexity’s AI browser shopping for you

The entire experience is powered by a version of GPT-5-mini that was trained specifically for shopping tasks, according to OpenAI. The company said that it was trained to «read trusted sites, cite reliable sources, and synthesize information across many sources,» as well as refine its prompts in real-time.

When compared to other ChatGPT models, such as GPT-5-Thinking or ChatGPT Search, shopping research leads in product accuracy. Yet, OpenAI acknowledged that it occasionally makes mistakes about product details, such as pricing and availability, and recommended that users always double-check its work.

How to use (and access it)

I found the experience of using it to be interactive and intuitive. To get started, all logged-in ChatGPT users, including those with Free, Go, Plus, and Pro plans, can either ask a shopping question, which will automatically activate the feature, or select the «Shopping research» option from the menu in the text box.

In your first prompt, describe what you want it to do for you. Then, ChatGPT will follow up with questions pertinent to your search, such as your budget or the features that are important to you. It will also use the context it knows about you, if you have those personalization toggles on, to tailor the response toward you.

As it conducts the research, it will display sample products it has found. With every product, you can indicate whether you are interested or not and why you made that decision, guiding the research further. This was my favorite part of using the feature, as it felt like an engaging Tinder-like experience, where you can quickly click through to indicate whether you like or dislike.

Then, after a few minutes, it will provide you with a personalized buyer’s guide that includes the top products, comparisons, and links that take you directly to the retailer’s website to place the order. In the future, the company plans to integrate this feature into the Instant Checkout experience, enabling you to make purchases directly on the site.

OpenAI said that user chats are never shared with retailers and that the results are generated organically, based on publicly available websites. Sites that want to appear in results must allow OpenAI’s crawlers to access their site, which can be done by following the instructions for the allowlisting process.

When I asked OpenAI how users can trust that the sites and products being recommended are from legitimate sources, I was told that the model was trained to avoid low-quality and spam sites, to help ensure users get the best user experience. At the same event, I had the opportunity to try the feature myself, and I had a lot more fun than I anticipated.

My experience

I got a sneak peek at the new feature, and the biggest standout to me was how interactive the experience was.

For example, in one demo, the user uploaded an image of Jeremy Strong in Succession and asked where he could find a similar turtleneck, for a lower price. While you likely could have reverse image searched that on Google, this time the user was able to enter a quick prompt, go through the refining «not interested» or «more like this» process, and get a set of options.

In another instance, I wanted to test finding suitable products for my dog. I typed in the prompt, «I want to buy dog treats for a 15-pound Yorkie.» It then asked follow-up questions, such as «Preferred treat types?» with multiple-choice answers that read: «Crunchy biscuits, jerky bites, soft chews, and freeze-dried.»

Then it showed me the quick feedback experience, where I could indicate whether I was more or less interested in the products. I was then able to say «not interested» and provide reasoning, such as «too expensive» or «my dog didn’t like this brand,» which helped narrow down the options. While the shopping guide was ultimately not groundbreaking, I found it helpful and would have felt comfortable purchasing one of its recommended products, or at least using it as a starting point.

Then I told ChatGPT that I wanted a dress for a specific semi-formal occasion that is fitted, features a pop of color, and is around $100. Out of the options that it gave me, some were so good that I took a picture on my phone to reference later.

But my favorite use case was using it to find gifts for people. Gift-giving is something I take particular pride in and enjoy doing, so I wanted to see if the AI could outdo me. I built a complex prompt: «I am looking for a gift for my friend who loves reading, has a 12lb yorkie, moved to a new apartment, wants to travel more next year, and is obsessed with Mickey Mouse. She is 32 years old and lives in NYC.»

It asked which gift themes I liked, of which I selected all: «reading and books, dog accessories, home decor, and travel essentials.» Then it provided some fun options. Of course, because I entered so many different characteristics, some product suggestions missed the mark, such as Mickey-themed everything. However, some were thoughtful, with my favorite being a gift card to The Strand, a bookstore I happen to love in NYC, and perhaps I wouldn’t have known about it if I didn’t live here.

The verdict

Ultimately, do I think it will completely replace how I shop? No. This is mostly because I love the art of finding the perfect product online, whether it’s a gift for someone or something I need myself.

However, I do think it could be a helpful, time-saving tool, especially if I’m trying to stay within a specific budget.