My AI shopping assistants are stuck in the past | The VergeSkip to main contentThe homepageThe VergeThe Verge logo.The VergeThe Verge logo.TechReviewsScienceEntertainmentAIHamburger Navigation ButtonThe homepageThe VergeThe Verge logo.Hamburger Navigation ButtonNavigation DrawerThe VergeThe Verge logo.Login / Sign UpcloseCloseSearchTechExpandAmazonAppleFacebookGoogleMicrosoftSamsungBusinessCreatorsMobilePolicySecurityTransportationReviewsExpandLaptopsPhonesHeadphonesTabletsSmart HomeSmartwatchesSpeakersDronesScienceExpandSpaceEnergyEnvironmentHealthEntertainmentExpandGamesTV ShowsMoviesAudioAIVerge ShoppingExpandBuying GuidesDealsGift GuidesSee All ShoppingCarsExpandElectric CarsAutonomous CarsRide-sharingScootersOther TransportationFeaturesVideosExpandYouTubeTikTokInstagramPodcastsExpandDecoderThe VergecastVersion HistoryNewslettersExpandThe Verge DailyInstallerVerge DealsNotepadOptimizerRegulatorThe StepbackArchivesStoreSubscribeFacebookThreadsInstagramYoutubeRSSThe VergeThe Verge logo.My AI shopping assistants are stuck in the pastComments DrawerCommentsLoading commentsGetting the conversation ready...AICloseAIPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All AIReportCloseReportPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All ReportTechCloseTechPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All TechMy AI shopping assistants are stuck in the pastI tried shopping with AI and almost got the wrong smartwatch.I tried shopping with AI and almost got the wrong smartwatch.by Stevie BonifieldCloseStevie BonifieldNews WriterPosts from this author will be added to your daily email digest and your homepage feed.FollowFollowSee All by Stevie BonifieldNov 27, 2025, 12:30 PM UTCLinkShareImage: The Verge, Getty ImagesStevie BonifieldCloseStevie BonifieldPosts from this author will be added to your daily email digest and your homepage feed.FollowFollowSee All by Stevie Bonifield is a news writer covering all things consumer tech. Stevie started out at Laptop Mag writing news and reviews on hardware, gaming, and AI.All I wanted was a new smartwatch. I had a couple in mind, like the Google Pixel Watch 4 or the Garmin Vivoactive 6, but my AI shopping buddies seemed really convinced watches from a couple years ago were the way to go.Over the past month, OpenAI, Google, Perplexity, and Microsoft have all introduced new shopping features on their AI platforms, right in time for the holiday shopping season. ChatGPT can generate a personalized buying guide for you. Copilot will help you track prices. Gemini will even call local stores for you.After asking four different AI bots to help me find a new watch, I’m more impressed than I expected, but also pretty disappointed. Each AI got a few things right, but the hiccups are both funny and crippling.I asked ChatGPT, Gemini, Perplexity, and Copilot the same question to start off my search: “Can you help me find a good Android smartwatch for my Nothing CMF Phone 1?” From there, I let each AI guide me through the research and shopping process.ChatGPT was the most conversational, and its new Shopping Research feature was the most involved and in-depth of the four AI models. It asked me what features mattered most to me (design and battery life) and showed me a dozen or so products to rate (“more/less like this”), then compiled a list of recommendations complete with a specs comparison chart. Afterward, it gave me a few links to deals on the top watch it suggested, the Garmin Vivoactive 5. The whole process took around 10 minutes.Screenshot: OpenAIThe Fitbit Versa 4, Google Pixel Watch 3, and Ticwatch Pro 5 were also in ChatGPT’s recommendations for me. It was able to compare battery life, app compatibility, prices, and “best for” use cases for each, like “daily use,” “budget friendly,” or “strong Android integration.”RelatedI sent ChatGPT Agent out to shop for meAdobe data shows consumers are embracing AI chatbots over traditional searchI wouldn’t normally turn to AI for help shopping for tech, but I can see how a feature like this could be useful to someone who isn’t sure what they’re looking for or where to start. Unfortunately, if I had trusted ChatGPT’s suggestions, I could have ended up with a last-generation Garmin watch, instead of the more recent Garmin Vivoactive 6.To be fair to ChatGPT, I didn’t specify “latest,” and the Garmin Vivoactive 5 is still widely available at retailers like Amazon and Best Buy. However, the newer Vivoactive 6 has some advantages that could have been deal-breakers, like more storage, GPS improvements, and new features like Smart Alarm. You might still go with the Vivoactive 5, especially if you want to save some money, but it could still be important to know what you’re sacrificing or just have more options to compare to.After waiting 15 minutes, Google’s AI emailed me to let me know that every store it called doesn’t sell Garmin smartwatches. PerfectThis issue cropped up on all four AI models, to varying degrees. Gemini suggested the “latest model” of a few watches, including the CMF Watch Pro 3, but also gave me a comparison chart for three top recommendations, all of which were from 2023 and 2022, including the Google Pixel Watch 2, which has been updated a couple of times now. Is the Pixel Watch 2 still out there for less money than the newer Pixel Watch 4? Sure. But it also has less battery life, uses an older charging setup, has thicker bezels, runs on an older processor, and only comes in one size instead of two.Perplexity suggested the most recent Pixel Watch 4, but also pointed me toward the Samsung Galaxy Watch 4… from 2021. Things got really interesting in Perplexity’s shopping tab, where the “More Products” section included a pile of cheap off-brand watches and even a phone. Perplexity made it easiest to get to actual purchasing links, though, which took longer to arrive at on ChatGPT and Gemini. If I had wanted a $7 “Smart Watch with Bluetooth Call,” I’d be in business.PreviousNext1/5Screenshot: PerplexityCopilot immediately suggested the CMF Watch Pro 2, which is specifically designed to pair with my CMF Phone 1, but was left out of most of the recommendations the other AI models gave me. Unfortunately, Copilot still overlooked the more recent CMF Watch Pro 3. However, I found Copilot’s shopping sidebar the most useful. It included a price history, aggregated pros and cons from Amazon reviews, a few different purchasing links, and an option to get price tracking notifications.Each AI successfully pulled up the CMF Watch Pro 3 when I tweaked my question to, “What are the best current smartwatches for the Nothing CMF Phone 1?” However, they continued suggesting older smartwatches alongside the Watch Pro 3. Again, the CMF Watch Pro 2 is still available and might be the right choice for some people. The AI models might even be pulling up older watches simply because newer models have fewer reviews.Still, you wouldn’t necessarily know you might be missing out on a more recent iteration of any of these products unless you specifically ask the AI for the “current” or “latest” front-runners. In contrast, if you read one of our reviews or watch a comparison video made by a real person, they’ll explain why you might (or might not) want the latest version of a product. AI can sometimes get there if you dig a bit for it, but it’s pretty hit-or-miss.RelatedGoogle’s AI try-on imagines your feet in new shoesBill Gates’ daughter Phoebe launched a shopping appThen there’s Google’s “Call for me” feature, which lets you send robocalls to local stores to see if they have the product you’re looking for. I had to use the Google app on my phone to get to this feature (rather than Gemini), and add “near me” or “nearby” at the end of my search query. When you scroll down in the location-specific results, you should see an option to let Google call for you, after asking you to confirm the details for your location and product request.After waiting 15 minutes for Google to fire off those phone calls, it emailed me to let me know that every store it called doesn’t sell Garmin smartwatches. Perfect.All these AI shopping tools might sound helpful on paper, but in practice, they aren’t sticking the landing. They seem to be pulling from product data that’s a couple of years old at this point, while also confidently giving you recommendations. If you don’t know what you’re looking for, it could be easy to end up with an outdated product, or even just miss out on newer options the AI doesn’t realize exist.The issue of time is enough of a deal breaker here that it makes it tough to recommend using any of these AI shopping helpers, at least right now. A couple of them are on the right track, particularly ChatGPT and Copilot, but they’re not suggesting current top picks often enough to replace up-to-date buying guides written by real people. For now, I’ll do my own smartwatch research.Follow topics and authors from this story to see more like this in your personalized homepage feed and to receive email updates.Stevie BonifieldCloseStevie BonifieldNews WriterPosts from this author will be added to your daily email digest and your homepage feed.FollowFollowSee All by Stevie BonifieldAICloseAIPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All AIGoogleCloseGooglePosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All GoogleMicrosoftCloseMicrosoftPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All MicrosoftOpenAICloseOpenAIPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All OpenAIReportCloseReportPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All ReportTechCloseTechPosts from this topic will be added to your daily email digest and your homepage feed.FollowFollowSee All TechMost PopularMost PopularWyze’s new security camera watches your yard from inside your homeYou can play classic Nintendo games on these custom SNES-inspired Nike sneakersI’m officially done with YouTube KidsYou’re buying a Frame TV? It’s okay to cheap out a littleFirst there was nothing, then there was Hoto and FanttikThe Verge DailyA free daily digest of the news that matters most.Email (required)Sign UpBy submitting your email, you agree to our Terms and Privacy Notice. This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.Advertiser Content FromThis is the title for the native adMore in AIOpenAI denies liability in teen suicide lawsuit, cites ‘misuse’ of ChatGPTCharacter.AI launches Stories for teens after banning them from chatsChatGPT and Copilot are being booted out of WhatsAppDavid Sacks tried to kill state AI laws — and it blew up in his faceWarner Music Group partners with Suno to offer AI likenesses of its artistsPerplexity says its AI personal shopper ‘puts you first’OpenAI denies liability in teen suicide lawsuit, cites ‘misuse’ of ChatGPTRichard Lawler12:43 AM UTCCharacter.AI launches Stories for teens after banning them from chatsEmma RothNov 26ChatGPT and Copilot are being booted out of WhatsAppDominic PrestonNov 26David Sacks tried to kill state AI laws — and it blew up in his faceTina NguyenNov 25Warner Music Group partners with Suno to offer AI likenesses of its artistsEmma RothNov 25Perplexity says its AI personal shopper ‘puts you first’Elissa WelleNov 25Advertiser Content FromThis is the title for the native adTop StoriesTwo hours agoThe VPN panic is only getting started1:00 AM UTCStranger Things 5 begins with a frantic race to the finish lineNov 26First there was nothing, then there was Hoto and FanttikNov 26You’re buying a Frame TV? It’s okay to cheap out a littleNov 25The AI boom is based on a fundamental mistake1:19 AM UTCThe best Black Friday deals we’ve found on TVs, laptops, and moreThe VergeThe Verge logo.FacebookThreadsInstagramYoutubeRSSContactTip UsCommunity GuidelinesArchivesAboutEthics StatementHow We Rate and Review ProductsCookie SettingsTerms of UsePrivacy NoticeCookie PolicyLicensing FAQAccessibilityPlatform Status© 2025 Vox Media, LLC. All Rights Reserved

The rapidly evolving landscape of AI shopping assistants presents a complex and, at present, somewhat frustrating experience for consumers. According to Stevie Bonifield’s reporting for The Verge, these tools, including models from OpenAI, Google, Perplexity, and Microsoft, are struggling to keep pace with the latest product offerings. The core issue centers around the fact that these AI models are primarily drawing from product data that is often several years old. While the potential for AI to streamline the shopping process is undeniable, the current implementation frequently leads to recommendations for outdated or superseded products.

Bonfield’s investigation highlights the specific shortcomings of each AI. ChatGPT, in particular, demonstrated a conversational and in-depth research approach, capable of generating personalized buying guides and detailed comparisons. However, it fell short by suggesting a Garmin Vivoactive 5 smartwatch, a model superseded by newer iterations. Similarly, Google’s “Call for me” feature, designed to automatically inquire about product availability in local stores, ultimately proved ineffective, failing to connect with retailers offering the desired products. Perplexity’s shopping tab, while providing a comparative analysis, also leaned heavily on older models, such as the Samsung Galaxy Watch 4 from 2021. Microsoft’s Copilot offered a more targeted approach, focusing on the CMF Watch Pro 2, but its recommendations still frequently prioritized older models alongside newer options.

The reliance on outdated data isn't simply a matter of minor inconvenience; it significantly impacts the utility of these AI assistants. Consumers seeking the most current or advanced products risk being directed to solutions that no longer represent the best available choices. The issue is further compounded by the potential for these AI models to underrepresent newer products when they haven’t yet accumulated a significant volume of reviews, a factor that can influence consumer confidence.

While the underlying technology driving these shopping assistants – particularly the ability to process and analyze vast amounts of data – holds considerable promise, the current state reflects a critical gap in real-time product information. Until these systems can more reliably access and incorporate the latest product data, the practicality and value of using them as primary shopping tools remain questionable. Bonfield's account serves as a cautionary note, suggesting a significant degree of skepticism is warranted at this time.