Recommended Free Tools
OpenAI and Perplexity have already added shopping features to their AI products, but that does not automatically make specialized shopping startups obsolete. The large platforms have a major advantage in reach and checkout. Startups argue they can still win when shopping depends on specialized product data, personal taste, or a complex task such as styling an outfit or furnishing a room. Whether that argument holds depends on results—not on the presence of a chat interface.
The practical distinction is between broad product discovery and category-specific decision-making. A general assistant can help build a shortlist; a strong specialist may help decide what actually fits your needs. Neither should be treated as an independent guarantee of the best product or price.
What changed: AI shopping moved from chat answers toward checkout
OpenAI announced Shopping Research on November 24, 2025, and Perplexity introduced shopping recommendations and an in-chat purchase option around the same period. The shift was not simply that chatbots could name products. Both companies moved toward a fuller shopping journey: understanding a request, finding and comparing products, and—in supported cases—helping complete a purchase.
That makes the original claim that startups “aren’t sweating it” a strategic argument from selected founders, not proof that specialist companies are safe. The important question now is whether broad platforms can deliver trustworthy, repeatable recommendations in categories where context and judgment matter.
Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchWhat shoppers get from ChatGPT and Perplexity
| Capability | ChatGPT | Perplexity |
|---|---|---|
| Research and discovery | Ordinary shopping queries can return product suggestions and comparisons. Shopping Research is a deeper mode that asks follow-up questions and synthesizes product information, reviews, specifications, and trade-offs. | Shopping is connected to Perplexity’s research-and-answer experience, with product discovery, summaries, and recommendations across retailers. |
| Personalization | Shopping Research can ask about preferences such as brand, size, performance, comfort, style, and price. ChatGPT may use memory when enabled. | Perplexity describes personalized recommendations informed by prior interactions or saved context. |
| Purchase flow | Instant Checkout supports purchases from eligible merchants; OpenAI announced Etsy sellers and Shopify merchants as initial participants. | Instant Buy uses PayPal for supported purchases. Perplexity’s help material describes availability for U.S. users; check the current experience for eligibility. |
| Important caveat | OpenAI says the price in an initial result may come from the first merchant shown and may not be the lowest available. | Checkout and product availability are not universal; verify the seller, price, and current terms. |
OpenAI’s Shopping Research announcement said the feature was rolling out on mobile and web to logged-in Free, Go, Plus, and Pro users. Its then-announced holiday usage terms should not be assumed to remain in force. The help documentation explains how it can ask clarifying questions and use memory when enabled. These tools may save research time, but their outputs still need checking.
For checkout, OpenAI describes Instant Checkout and its Agentic Commerce Protocol. Shopify says its catalog can provide product information such as prices, inventory, images, and variants to AI shopping experiences through its OpenAI commerce work. Perplexity’s Instant Buy support page documents PayPal-powered checkout, but eligibility and offers can change. An integration does not mean every store or item can be bought in chat.
Why specialized shopping companies see an opening
Companies named in the 2025 coverage included fashion-focused Daydream and shopping startups Phia and Cherry, as well as Onton, which TechCrunch described as focused on interior-design discovery and formerly known as Deft. Their founders’ common thesis is that shopping is not always a search problem. A large model may be able to retrieve and summarize product pages; a specialist hopes to make a better decision for a particular kind of shopper.
That distinction has several layers:
- Information retrieval: Find items matching explicit constraints, such as a laptop with a certain screen size and budget.
- Decision support: Explain meaningful trade-offs among products that meet those constraints.
- Taste inference: Learn which styles, materials, or design choices a particular person is likely to prefer.
- Merchandising: Assemble a coherent outfit, room, routine, or compatible set rather than return unrelated individual products.
- Workflow completion: Help the shopper carry a project from initial idea through configuration, visualization, and purchase.
Product data can matter as much as model size. Catalogs often describe similar things inconsistently, omit useful attributes, or mix variants. A vertical company may normalize the details that matter within one category:
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errors- Fashion: silhouette, fabric, cut, occasion, fit, styling relationships, and brand identity.
- Furniture: dimensions, materials, color, style, room fit, and how pieces work together visually.
- Beauty: ingredients, formulation, finish, skin type, routine, and relevant sensitivities.
- Electronics: compatibility, ports, ecosystem, performance, repairability, and the buyer’s existing setup.
TechCrunch reported that Onton’s CEO described a pipeline that cataloged hundreds of thousands of interior-design products in a cleaner format for its internal models. Daydream’s Julie Bornstein highlighted fashion concepts such as silhouette, fabric, occasion, and how people build outfits over time. Those are examples of the specialist thesis, not independently verified evidence that these products outperform ChatGPT or Perplexity. The original reporting is at TechCrunch.
Where general-purpose assistants have the advantage
ChatGPT and Perplexity start with something many startups must work hard to earn: an existing relationship with users. Someone already using an assistant for research or planning may try its shopping feature without downloading another app or learning a new service. The same platform can also cover many categories, invest in models and ranking systems, and negotiate merchant, product-data, or payment integrations.
Rank #3
That breadth is especially useful when the decision is mostly about comparing known specifications. A shopper can ask for a shortlist of laptops under a set budget, compare appliance features, or find luggage with stated dimensions. The assistant can organize options quickly, and broad coverage may matter more than deep taste modeling.
OpenAI says its merchant ranking considers factors including availability, price, quality, primary-seller status, and whether Instant Checkout is enabled. It also says merchants pay a fee on completed purchases and that this fee does not affect results. That is OpenAI’s stated policy, not an independently established finding about every recommendation. See its commerce announcement for the company’s description.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Where a vertical product could be better
A specialist has a stronger case when the shopper’s real need is not “show me products like this,” but “help me choose something that works with my life.” Which coat fits an existing wardrobe, personal preferences, and local climate? Which table fits a room’s dimensions and visual style? Which device works with equipment the buyer already owns? Those questions require context across products, constraints, and sometimes images—not just a larger list of results.
Vertical businesses can also own a useful workflow: outfit construction rather than single-item search, room visualization rather than furniture lookup, a compatible technology setup rather than isolated specifications, or expert and community curation rather than a generic ranking. That can create value even if a large platform remains the shopper’s first stop. A specialist might provide the catalog intelligence or decision layer behind the discovery and checkout journey.
But “vertical” is not itself a moat. A narrow brand voice or chat window can be copied. Better data is valuable only if it is difficult to reproduce, kept current, and demonstrably improves decisions, conversion, retention, or merchant economics. Large platforms can license data, add partnerships, improve product feeds, or build category-specific tools.
The business-model and trust problem
Shopping recommendations sit close to money, so the way a service earns revenue matters. Possible models include affiliate referrals, merchant transaction fees, sponsored placements, subscriptions, or software and catalog services for retailers. Each can support a product, but it can also create incentives shoppers should understand.
Users should be able to tell whether a placement is sponsored, whether the company may earn from a referral or sale, and whether checkout eligibility affects what appears. Merchants and platforms should make ranking policies intelligible rather than relying on a vague promise of neutrality. Even when a company says payment does not influence rankings, the shopper still needs clear disclosure and a way to compare options.
There are ordinary accuracy risks, too: a price or stock level can become stale; a model can confuse similar versions; a review summary can blur firsthand testing and customer anecdotes; shipping, warranty, and return terms may be missing. In beauty, child-safety, electrical, health-related, and specialist equipment categories, a plausible-sounding recommendation is not a substitute for qualified guidance or checking manufacturer requirements.
The ShoppingComp benchmark reported critical errors by shopping agents, including unsafe-use recommendations and misinformation about promotions. It is useful evidence that agentic shopping has failure modes, not a direct test proving that every currently deployed ChatGPT or Perplexity feature makes the same errors.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.How to judge whether the startups’ thesis is working
Claims about proprietary data or better personalization are not enough. A specialist needs to show that its advantage changes outcomes. Useful evidence would include:
- Recommendation quality: Does it satisfy explicit constraints, avoid unsuitable options, explain trade-offs correctly, and offer coherent alternatives?
- Preference fit: Does it learn useful preferences rather than repeatedly asking the same questions or overfitting to a few past choices?
- Conversion and retention: Do users complete purchases and return for their next decision?
- Returns and satisfaction: Do recommendations reduce avoidable returns or leave shoppers more satisfied?
- Time to decision: Does the product shorten the process without simply hiding important trade-offs?
- Merchant value: Does it send qualified buyers, improve revenue per visitor, or reduce the cost of product discovery?
- Trust: Are ads, affiliate relationships, review sources, price freshness, and ranking incentives clearly disclosed?
The 2025 TechCrunch story offered founder arguments, not controlled head-to-head tests establishing that Onton, Daydream, Phia, Cherry, or another specialist performs better than the large platforms. Investors and operators should treat the specialist advantage as a hypothesis to measure, not a settled fact.
What shoppers should do now
- Use a general assistant to map the market. Ask for options that fit your budget and stated constraints, then request the trade-offs and the assumptions behind the shortlist.
- Use a specialist when the category demands judgment. Fashion, interiors, beauty, and compatibility-heavy purchases may benefit from tools that understand the relevant attributes or workflow. Compare the specialist’s actual output with a general assistant rather than assuming the niche label guarantees quality.
- Check the exact item before paying. Confirm model number, size or variant, current price, stock, seller identity, condition, warranty, shipping deadline, and return policy on the merchant’s page. OpenAI specifically warns that a price in an initial shopping result may not be the lowest available: its shopping-search guidance explains the caveat.
- Verify high-stakes claims independently. For safety-sensitive, health-related, expensive, or hard-to-return purchases, check manufacturer specifications, qualified advice, and more than one source. Treat a review summary as a pointer to evidence, not as evidence of testing by the assistant.
- Notice the incentives. Look for sponsorship and affiliate disclosures, and ask whether the assistant is showing a broad selection or only eligible merchants and checkout partners.
The likely outcome: coexistence, with pressure on shallow products
The likely competitive split is not simply “big AI companies versus startups.” General assistants are well positioned to become broad discovery and comparison front doors. Vertical products can remain useful where specialist data, taste, visualization, workflow, or expertise makes the decision materially better. A company offering only a generic chat interface is more exposed than one whose product improves the underlying catalog, the quality of choices, or the transaction outcome.
For consumers, the best tool is the one that reduces effort without concealing uncertainty or incentives. For a startup, the test is whether its specialization creates better outcomes people and merchants can measure—not whether its founders feel calm about a platform launch.
For merchants, making accurate product information available to AI channels may expand discovery, but it does not guarantee placement or sales. Shopify’s agentic storefront documentation describes the merchant-side setup and limitations; clean product data, current inventory, accurate variants, and reliable fulfillment remain essential.
The Tool Desk
Outbyte Driver Updater FREEFix the driver behind crashes, sound loss and screen glitchesFind Drivers →Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
