Amazon’s Buy for Me Experiment Exposes the Dark Side of Agentic Commerce

Amazon’s latest experiment in AI-driven shopping – a feature called “Buy for Me” – is revealing a troubling side of agentic commerce. This feature allows Amazon’s AI to do more than just recommend products; it actually purchases items on customers’ behalf from other brands’ websites. On the surface, it seems convenient: shoppers can discover and buy items from independent brands without ever leaving Amazon’s app. But for the brands whose products are suddenly showing up on Amazon without permission, Buy for Me has become a wake-up call. By scraping public product data, auto-generating Amazon listings, and acting as an intermediary buyer, Amazon is testing a model of AI-driven commerce that puts platform control above merchant consent. This raises urgent questions about who controls product information, who “owns” the customer relationship, and what rights a platform has to execute sales in the age of AI shopping agents.

What Is Amazon Buy For Me and How Does It Work?

Buy for Me is part of Amazon’s BuyForMe program, an AI-powered shopping feature within Amazon’s app that Amazon began piloting in 2025. Amazon’s Buy for Me is currently in beta and available to a subset of U.S. customers using the Amazon Shopping app on iOS and Android. It allows Amazon users to purchase products sold on a brand’s website, directly through the Amazon interface. Amazon’s AI powers this feature, automating the process of purchasing from a brand’s website within Amazon’s app. In practice, Amazon’s system finds products that are not sold on Amazon’s marketplace but are available on independent brand sites (for example, a small Shopify-powered store). It then creates a listing on Amazon’s store for those products, labelled as coming from “other brands.” Shoppers might see these listings mixed into their Amazon search results with a special “Buy for Me” button. Branded items from other stores and shop brand sites directly can be found via the search bar, and these are shown in a separate section of relevant results. Importantly, these are not third-party sellers who signed up for Amazon – they are automatically added by Amazon’s AI as part of Amazon’s shopping experience within Amazon’s app. Amazon’s shopping experience is expanded by integrating external brand websites into Amazon’s store, allowing customers to purchase products from other sites without leaving Amazon’s app.

From the shopper’s perspective, using Buy for Me feels similar to a normal Amazon purchase. They can add the product to their Amazon cart and check out using their Amazon account, without visiting the brand’s own site. However, the item isn’t stocked or shipped by Amazon. Behind the scenes, Amazon’s AI assistant acts as a go-between: it takes the order details and, acting on the customer’s behalf, places an order on the brand’s website. Before creating a listing, Amazon’s system checks product and pricing information on the brand’s website to ensure accuracy. Amazon’s AI securely transmits the customer’s encrypted personal and payment details to the brand’s website to complete the transaction. Essentially, Amazon itself becomes a “customer” of the independent merchant, executing the purchase with the customer’s information (which Amazon securely provides from the user’s saved details). Once Amazon completes the purchase on a customer’s behalf, the customer receives an auto generated email (order confirmation) from the brand store. The merchant then fulfills the order and ships it directly to the Amazon shopper. Delivery, returns, and customer service for purchases made through Buy for Me are managed by the brand store, not Amazon, and some merchants may use Shopify shipping notification emails for order updates.

In simpler terms: Amazon’s Buy for Me lets customers purchase products on Amazon for an item that Amazon doesn’t sell. Amazon’s system will buy it from the brand’s site for you, so you never have to leave Amazon. Customers can link directly to brand’s websites or purchase items from shop brand sites through Amazon’s Buy for Me feature. The checkout process includes applicable taxes, and Amazon does not charge a commission for purchases made through Buy for Me during its beta phase. The orders tab in Amazon’s app allows customers to track these purchases, but separate orders from different brands or stores are not displayed together. The appeal for users is clear – one-stop shopping and Amazon’s checkout convenience applied to almost any product on the web. Amazon even extends its customer protections (like its A-to-Z guarantee and unified order tracking) to these purchases. For Amazon, it keeps customers inside the Amazon ecosystem and potentially expands product selection infinitely by tapping into other retailers’ catalogs. Amazon plans to expand the Buy for Me feature to more customers and brands over time, further increasing the reach of Amazon’s shopping experience. But for the merchants whose products are being bought “for” customers by Amazon, the process is anything but straightforward.

An AI Middleman in Action

To make Buy for Me possible, Amazon employs what it calls “agentic AI capabilities”. Agentic commerce refers to autonomous AI agents that act independently on behalf of customers to accomplish shopping goals, going beyond traditional AI by making decisions and completing transactions without user intervention. This means the AI isn’t just answering questions – it’s taking actions online. The AI scours a brand’s public website for product and pricing information, likely using web crawling or integrations, and generates a product listing on Amazon based on that data. Amazon’s system checks verify product information and stock status by cross-referencing the brand’s website data before displaying or updating listings. It will periodically check the brand’s site for price changes or stock availability so it can update the Amazon listing. (However, as many merchants discovered, this process isn’t perfect – more on that below.) When an Amazon customer clicks “Buy for Me,” the AI proceeds to simulate a customer checkout on the brand’s site:

• It adds the item to the website’s cart, just as a regular shopper would.
• It uses the Amazon-held payment details and shipping address of the customer to fill in the order form. (Amazon has stated that it encrypts and securely transmits this info, so the merchant never sees the actual credit card numbers – they simply receive a normal order paid via a card.)
• The order is placed on the merchant’s website, with a unique Amazon-generated email address (something like xyz123@buyforme.amazon.com) as the contact. This allows Amazon to monitor the order status and handle communication, while shielding the customer’s personal email.

After this, the merchant’s own system processes the order. From the merchant’s viewpoint, an order from a customer has appeared out of nowhere – often flagged with that strange @buyforme.amazon.com email. The merchant will pack and ship the product to the address provided (which is the real customer’s address). Amazon typically sends the customer shipping updates through its app or email, and if the customer has an issue or wants to return the item, Amazon facilitates that (often by providing return labels or support via its customer service). In effect, Amazon acts as an agent and a buffer: the customer still goes through Amazon for service, and the merchant is order fulfillment that Amazon routed to them. fulfilling an order

Crucially, all of this happens without the merchant ever having listed their product on Amazon themselves. The listings are auto-generated by Amazon’s AI; the merchant didn’t write the title or description on Amazon, didn’t set up an Amazon seller account, and didn’t explicitly agree to sell on Amazon’s marketplace. This is unlike any traditional Amazon marketplace transaction, where the seller actively participates. Buy for Me blurs the line – the merchant becomes an unwitting drop-shipper fulfilling an Amazon-placed order. Notably, Buy for Me is currently a beta program and is still in testing, which has led to issues for small businesses regarding control and potential legal risks.

Listed Without Consent: A Marketplace Without Independent Sellers?

The most controversial aspect of Buy for Me is that it enrolls merchants by default – without asking. Merchants discovered that Amazon had created listings for their products in Amazon’s ecosystem even though they never agreed to sell on Amazon. It’s essentially Amazon “listing” products from other sites as if they were part of Amazon’s catalog, then buying them on behalf of customers. This program might be seen as part of Amazon’s long-running ambition to be the place where customers start and finish their shopping journey – even if the actual transaction happens elsewhere.

According to merchants’ accounts, this happened quietly. Brands started noticing orders coming in through their own ecommerce sites with a strange email address domain (something like buyforme.amazon.com), and discovered their products were appearing on Amazon with a “Buy for Me” button. Many merchants only found out because customers reached out confused, or because unexpected orders started coming in. Amazon did not notify them ahead of time or ask permission. As for compensation, merchants do receive payment for any orders Amazon places on their site (just like a regular customer sale, minus whatever payment processing fees they normally pay). However, Amazon does not take a marketplace commission on top – it’s not like a 15% fee as in a typical Amazon sale. Amazon’s “gain” is keeping the customer on its platform and potentially earning their loyalty (and capturing data). The merchant gets the revenue from the product sale, but they didn’t explicitly agree to Amazon being a sales channel.

It’s easy to see why Amazon would frame this as helpful. Amazon can say: “We’re bringing you new customers.” And in some cases, merchants did see a small lift in sales. But the broader concern is the precedent: if Amazon can unilaterally add products from any website into Amazon’s shopping results, it effectively treats the open web as its marketplace inventory. The merchant loses control of where and how their products are marketed and sold.

This is also a troubling inversion of the traditional relationship between platforms and sellers. Usually, sellers opt into a marketplace and agree to its rules, fees, and customer relationship structure. Here, the platform opted the seller in, with the “rule” being: if your site is publicly selling something, Amazon might list it and facilitate the sale. That’s a radical shift in how consent works in ecommerce.

When Good Intentions Go Wrong: Merchant Outrage and Real Problems

The lack of consent is a principle issue, but equally important are the practical problems that arose from these unauthorized listings. By acting on second-hand data and automating purchases, Amazon’s AI introduced errors and confusion that merchants had to clean up:

• Out-of-stock items and outdated info: Because the AI scraped product info at some point in time, it sometimes listed products that the brand no longer had available. Customers placed orders on Amazon for items that didn’t exist in the merchant’s inventory. This led to merchants scrambling to cancel orders or explain to angry buyers that the product was unavailable. The very first clues many got about Buy for Me were these unexpected orders for long-gone products. Often, these orders arrived via an auto generated email address created by Amazon, which made it difficult for merchants to immediately recognize or verify the legitimacy of the order.
• Mismatched products and descriptions: Some merchants reported that Amazon’s auto-generated listings didn’t always match the product perfectly. In one case, a customer thought they were buying a large version of a stress-ball toy (based on Amazon’s listing), but the merchant only sold a smaller size – and that’s what was shipped. The AI had apparently misinterpreted or merged product data, resulting in the wrong item being delivered. The customer blamed the small business for “sending the wrong product,” hurting the brand’s reputation through no fault of their own.
• Incorrect pricing or terms: A few merchants saw Amazon display prices that didn’t match their current pricing – potentially a caching issue or a misunderstanding like showing a wholesale/bulk price or an old sale price. This could mean customers were charged a different amount than the product actually costs on the site, leading to confusion and potential loss for someone (either the customer pays more than they should, or the merchant has to decide whether to honor a lower price they never set on Amazon).
• Customer confusion over who they bought from: Several merchants noted that customers thought they had ordered from Amazon directly. The Amazon-generated product pages, while labeled as from “other brands,” still looked like typical Amazon pages to many shoppers. So when an issue arose – wrong item, delayed shipment, etc. – some buyers contacted Amazon support expecting resolution, while others contacted the merchant (since the package ultimately came from the merchant’s warehouse). Small businesses suddenly found themselves fielding customer service issues caused by Amazon’s system, often having to explain, “We didn’t list our product on Amazon; Amazon’s AI did this.” This scenario put brand trust at risk. A customer who has a bad experience might leave a negative review or lose faith in the brand, not realizing the disconnect in the sales channel. Additionally, some merchants use Shopify shipping notification emails to communicate with customers, but when orders are placed via Amazon’s Buy for Me, this can cause confusion—customers may receive both Amazon and Shopify notifications, making it unclear who is responsible for the order and shipment.
• Returns and fulfillment burden: Because the orders are fulfilled by the merchants, any returns or exchanges fall to them as well. One major headache was that if Amazon’s info was wrong (say, the wrong size was listed) and the customer wanted a return, the small merchant had to handle the return shipping and refund. Amazon wasn’t automatically compensating these errors; in effect, the merchant eats the cost or inconvenience, unless they escalate a complaint to Amazon. Some merchants reported offering refunds or replacements to appease customers, essentially cleaning up the AI’s mistakes. Offering free return labels in these situations can help mitigate disputes, improve customer satisfaction, and reduce the risk of chargebacks, but it also adds to the merchant’s operational costs.
• Operational strain and inventory management: A few artisan or very small-scale sellers worried, what if this took off suddenly? If Amazon’s algorithm decided to push their product and they got a spike of orders, could they even handle it? One jewelry maker said, “If suddenly there were 100 orders, I couldn’t necessarily manage… I should be asked about that. This is my business.” For micro-businesses, being unknowingly featured on the world’s largest store is a stress-test they never signed up for.
• Policy and partnership conflicts: At least one merchant pointed out that they carry other independent brands’ products in their store, and some of those brands explicitly forbid selling on Amazon (to maintain exclusivity or brand positioning). By Amazon pulling those products onto its site via this merchant’s catalog, it could put the merchant in breach of agreements with their partners. Others mentioned the unauthorized use of their product photography and descriptions (often copyrighted content) by Amazon’s listings, raising intellectual property concerns. It felt like Amazon assumed anything publicly visible online was free to reuse commercially.

All these issues fuel the outrage, but the prevailing sentiment from merchants is less about any single order gone wrong and more about loss of control. These entrepreneurs carefully cultivate their brand image, customer experience, and sales channels. Suddenly they woke up to find their brand presented on Amazon in a way they didn’t choose, with content they didn’t vet, and funneling orders in a manner that cut them out of the loop. Even those who initially saw extra orders roll in (and thought “hey, new sales!”) quickly grew wary when errors and complaints surfaced. As one affected seller said, “When things started to go wrong, there was no system set up by Amazon to resolve it. It’s just: We set this up for you, you should be grateful… now you deal with it.” That feeling of powerlessness – that Amazon can reach into their business and meddle with how products are sold – is what really underlines the “dark side” of this agentic commerce experiment.

Execution Without Consent: A Dangerous Precedent

The bigger story here is that Buy for Me exposes a fault line in ecommerce: the ability to execute a transaction has become decoupled from permission. In other words, an AI can now act on public-facing information to accomplish shopping goals, including making decisions and completing transactions without user intervention. Normally, if a platform wants to sell a product, it needs some agreement with the merchant or brand (be it a wholesale agreement, a marketplace seller contract, or at least an affiliate marketing consent). Here, Amazon effectively said: “If your website is publicly selling item X, our AI can offer item X to our users and facilitate buying it – no explicit consent required from you.” In Amazon’s view, the open web becomes its supply catalog. The Buy for Me feature is integrated within Amazon’s store, allowing the AI-powered tool to facilitate cross-store product discovery, purchase, and even product scraping from third-party brands’ websites, often without the explicit consent of small businesses. This has led to concerns over control, transparency, and potential legal issues.

This raises several profound questions and challenges assumptions in the ecommerce world:

• Who owns the “product truth”? When Amazon scrapes a product page and presents that information on its own site, it becomes a secondary source of truth. As we saw, the data can be wrong or stale, meaning Amazon’s version of the product might differ from the reality on the merchant’s site. The merchant loses control over how their product is described and portrayed at the point of discovery. If Amazon’s listing has an incorrect photo or missing info, it might lead to a bad customer experience – yet the merchant had no hand in creating that content. In the age of AI, if agents are pulling info from everywhere, maintaining accurate product data integrity is a challenge. Merchants are rightly asking: Shouldn’t we have control over our own product narrative? With Buy for Me, that control was ceded to Amazon’s algorithms.
• Who owns the customer relationship? In a Buy for Me transaction, the customer is an Amazon customer first and foremost. The shopper likely thinks of it as an Amazon purchase, and they may not even be aware of the brand’s standalone store. Amazon intermediates the entire experience: the product discovery happened on Amazon, the checkout was on Amazon, and any post-purchase communication may come via Amazon. The merchant, while fulfilling the order, doesn’t get the usual benefits of a direct sale. For instance, they get an order with a name and address – but the email is an Amazon proxy address, so they can’t add that customer to their mailing list or easily follow up with marketing. They also know nothing about that customer’s browsing behavior or history like they would if the customer shopped on their site. In essence, Amazon has inserted itself between the merchant and the end-buyer, preserving all the customer data and loyalty for itself. The merchant gets a sale, but loses a potential long-term customer, because that buyer may never actually engage with the brand beyond Amazon. Importantly, Amazon does not access or reveal customers’ previous orders from other brands’ websites, which highlights privacy and security aspects related to customers’ previous shopping history.
• Who has the right to execute a sale? Perhaps the most unsettling question: can a platform like Amazon initiate a sales transaction for a product it doesn’t own or officially resell? The merchants’ stance is that execution of a sale – taking an order and payment – should be under the merchant’s control (or their chosen retail partners). Amazon flipped this around by executing sales on the merchant’s behalf without asking. It’s a bit like someone deciding to open a storefront for your products without asking, because they found your catalog in a brochure. In the AI era, the “someone” doing that can be a powerful algorithm operating at massive scale. Discovery is collapsing into action: an AI can move from finding a product online to completing the checkout for a user in seconds. If this becomes the norm, merchants might have to grapple with a world where any published product info is an API for purchase. That broken assumption – “if I didn’t explicitly list it on a platform, it can’t be sold there” – is what Buy for Me shattered.

Within Amazon’s store, customers search for products as usual, but now the ecosystem includes external brand listings, expanding the scope of product discovery and purchase options.

To be clear, Amazon isn’t doing this out of malice toward small businesses; it’s doing it to remove friction for shoppers (and to keep them from going elsewhere). But the asymmetry of power is stark. Amazon decided the new rules unilaterally, leaving merchants to react. It highlights an uncomfortable truth: in the absence of explicit protections or standards, tech giants may leverage AI to route around traditional consent in commerce. Today it’s Amazon with Buy for Me; tomorrow it could be other large platforms finding ways to leverage publicly available data to their advantage.

The Universal Commerce Protocol: Consent-Based Agentic Commerce

Is there a better way to harness AI in shopping without trampling on merchant consent? Many in the industry believe so, and they’re rallying around an alternative approach called the Universal Commerce Protocol (UCP). UCP is a newly introduced open standard designed specifically for agentic commerce, but unlike Amazon’s closed experiment, UCP is built on explicit, machine-readable consent from merchants.

Under the Universal Commerce Protocol, merchants voluntarily expose their product data and purchase workflows via a standardized API or manifest. In plain terms, a brand can signal to AI agents: “Here’s how you can work with my store if you want to buy something.” This manifest includes real-time product details (pricing, stock, descriptions), rules for checkout (available shipping methods, tax calculations, promo codes, etc.), and how to actually submit an order and payment. Because it’s machine-readable and standardized, any AI shopping assistant that speaks UCP can understand and transact with the store only in the ways the merchant allows.

Several big names are backing UCP – it was co-developed by Google along with partners like Shopify, Walmart, Target, and others. The reason is clear: they envision a future where AI shopping agents become common, and they want a level playing field where retailers have control and buyers have choice. In a UCP scenario, if a shopper asks an AI assistant (say Google’s chatbot or some voice assistant) to buy a product, the assistant would search for merchants that support UCP for that product. It could perhaps find multiple options and compare prices or loyalty benefits. When it goes to execute the purchase, it would use the UCP interface to do so seamlessly. Importantly:

• The merchant remains the “seller of record”. The sale happens as if on the merchant’s site (just automated). The merchant sets the terms of sale, and they know an AI agent is checking out under a real customer’s authorization.
  
• The merchant likely gets to retain the customer relationship (for example, the protocol could allow the real customer email to be shared in a secure way, or at least not hide the brand behind an alias).
  
• Because the data comes directly from the merchant’s feed, the product info is accurate and up-to-date. The AI doesn’t have to scrape webpages and risk errors; it’s getting official data.
  
• If a merchant doesn’t want certain products sold via third-party agents or has certain conditions (like “don’t allow discount codes beyond X” or “limit 2 per customer”), those rules can be encoded in the protocol. The AI must respect those rules to complete the purchase successfully.
  
• In short, consent and control are baked in. Merchants opt in to UCP and thereby agree to let participating AI agents facilitate sales under agreed-upon rules. If they opt out, the AI should leave them alone.
  

It’s a very different philosophy from Amazon’s Buy for Me. One is “Let’s collaborate via open standards”, the other is “We’ll do it anyway, try to stop us.” UCP is still brand new (announced in early 2026), and Amazon was notably absent from its supporters. That’s not surprising – Amazon typically prefers its own closed ecosystem. In fact, while Walmart and Target jumped on the UCP bandwagon (signaling their interest in being more open), Amazon has shown no sign of adopting UCP or similar standards. Instead, Amazon has been building features like Buy for Me and its AI assistant (nicknamed “Rufus” internally) to strengthen its walled garden.

Consent vs. Power: Two Visions for AI Shopping

We are essentially seeing a fork in the road for AI-driven commerce:

On one side is the consent-based model – exemplified by UCP and the stance of companies like Google, Shopify, and Walmart. This vision treats merchant permission as paramount. It aims for a future where your personal AI shopper can roam the web and buy from any store that has chosen to welcome such agents. It’s a world of interoperability, where open standards enable any retailer to plug in and any AI to transact with minimal friction, all while respecting each retailer’s boundaries. In this model, merchants and agents work together: a merchant says “I’m open for AI-driven business” and an agent says “great, I’ll bring you a customer and follow your rules.”

On the other side is the power-based model – typified by Amazon’s approach. This vision is about leveraging platform power to make shopping as frictionless as possible for consumers, even if that means acting without explicit partner approval. It’s not that Amazon is against interoperability in principle, but rather that Amazon prioritizes speed, control, and keeping customers inside its own ecosystem. In this model, the platform’s AI might act around merchants rather than with them. If the data is out there and the platform has the capability, it will use it to serve the customer (and itself), counting on its dominance to mitigate pushback. Amazon’s stance seems to be: customer convenience first, we’ll sort out the messy details later. After all, Amazon likely believes that if it can deliver sales to merchants, most will eventually come around or at least tolerate it.

This divide is not just technical; it’s philosophical. It touches on trust and the balance of power in digital commerce. Many independent businesses and industry watchers are alarmed that if a company as influential as Amazon normalizes “execution without consent,” it could set a precedent. Smaller platforms might be pressured to follow suit just to compete, scraping and transacting without merchant agreements. The result could be a race where merchants have even less control, and only large aggregators hold the keys to customer access.

However, the pushback and publicity around Buy for Me suggest another outcome is possible. The outcry from merchants is a signal that consent still matters. It’s forcing a conversation in the industry: Will the future of shopping be opt-in and cooperative, or opt-out and dictated by the biggest player? Early signs of an “open vs closed” battle are here – one commentator dubbed it an “anti-Amazon alliance” forming (Google, major retailers, and payment providers aligning on standards like UCP, essentially everyone except Amazon teaming up).

Even Amazon’s own CEO has acknowledged agentic commerce is coming in a big way, though he’s publicly skeptical of third-party agents’ quality right now (pointing out that many AI shopping tools often show wrong info). Amazon’s answer to that seems to be: we’ll build our own AI shopping experience (like Rufus and Buy for Me) to ensure the quality and keep control. In contrast, competitors are saying: we’ll make it easy for any AI to shop anywhere, so no single platform controls everything. It’s a clash between building a closed network effect versus embracing a federated ecosystem.

New Reality for Merchants: Product and Pricing Information as Open Invitations

For merchants – especially those direct-to-consumer brands and sellers who have relied on independence – the Buy for Me experiment is a wake-up call about a broken assumption. Many assumed that if they didn’t list products on a marketplace like Amazon, those products simply wouldn’t be sold there. Going forward, that assumption may no longer hold. If your product data is publicly accessible, AI agents may increasingly treat it as “executable.” In other words, any information on your site (product name, price, “Add to Cart” button) could be used by an agent to initiate a purchase, with or without asking you.

This doesn’t mean every merchant should suddenly hide their data or lock down their site. But it does mean the industry needs to catch up in terms of norms and maybe protections. We’re entering an era where the lines between a platform and the open web are blurring. In the same way Google indexed the world’s information (sometimes to the chagrin of content owners), AI shopping tools might index and leverage the world’s products. The critical difference is that this isn’t just showing an image or a snippet – it’s actually completing transactions.

For now, Amazon’s Buy for Me is an extreme example and it has faced significant backlash. Amazon may adjust how it operates or which products it includes, especially as merchants opt out. But even if Amazon treads more carefully, the concept is out in the wild. Merchants should recognize that discovery can instantly turn into purchase in the AI age. The best outcome would be that consent-based systems like UCP gain traction, so merchants can explicitly decide how their data is used by shopping agents. The worst outcome would be that merchants have to constantly play defense, reacting to one platform’s unilateral moves after another.

In the immediate term, brands caught off guard by Buy for Me have learned to be vigilant. Many are searching Amazon proactively to see if their catalog is listed, and sharing information on how to opt out or shut down unwanted “ghost listings.” It’s a reactive solution, though, and it underscores the power imbalance: Amazon didn’t ask first, but now each merchant must ask Amazon to remove them if they don’t want to participate.

Agentic commerce – AI agents shopping on our behalf – is clearly on the horizon, and it promises great convenience. But Amazon’s experiment shows the darker side of what can happen when those agents operate purely on the terms of the platform deploying them. Going forward, the hope among many in the ecommerce community is that consent, accuracy, and fairness will become core pillars of agentic commerce, rather than afterthoughts. Otherwise, we may see more flashpoints like Buy for Me where innovation collides with the rights and relationships of those who make and sell the products.

The Technology Behind Buy for Me

At the heart of Amazon’s Buy for Me feature is a sophisticated blend of agentic AI capabilities, designed to make online shopping more seamless than ever. Unlike traditional AI, which might simply recommend products or answer questions, agentic AI in the Amazon shopping app can actually take action on a customer’s behalf—navigating other brands’ sites, filling in checkout forms, and completing purchases without the customer ever leaving Amazon.

Here’s how it works: When a customer finds a product through Buy for Me, Amazon’s AI agent springs into action. It gathers up-to-date product and pricing information directly from the brand’s website, with this data refreshed regularly to ensure accuracy. Once the customer decides to buy, the AI agent securely provides the customer’s encrypted name, address, and payment details to the independent seller’s checkout page. This process is designed to protect customer data, as Amazon acts as an intermediary—brands receive only the information necessary to fulfill the order, while sensitive payment details remain shielded.

A unique relay email address is generated for each order, allowing for efficient communication between the customer and the brand without exposing personal contact information. This not only streamlines order updates and support but also helps maintain privacy and security throughout the transaction.

The technology also ensures that product availability and pricing information are checked by Amazon’s system before the purchase is completed, reducing the risk of ordering out-of-stock items or encountering price discrepancies. For sellers seeking alternatives to Fulfillment By Amazon (FBA), options like third-party logistics providers can help manage fulfillment complexity as new AI-driven sales channels emerge.

Frequently Asked Questions

Does Amazon Buy for Me mean Amazon is dropshipping products from other stores?

Not exactly in the traditional sense, but it’s close. In dropshipping, a seller lists products they don’t stock and then buys them from a supplier when ordered. With Buy for Me, Amazon is listing products it doesn’t stock and then purchasing them from the brand’s site when a customer orders. The difference is Amazon isn’t doing it to profit on product margin (at least not directly, since it isn’t charging fees). It’s doing it to keep the customer in Amazon’s shopping experience. From the merchant’s perspective, though, it can feel like being used as an unconsented dropshipping supplier for Amazon’s marketplace interface.

Is Amazon paying merchants for being listed in Buy for Me?

No. Merchants are not being paid for inclusion. They only get paid if an actual order is placed, in which case they receive payment like any other direct online sale through their own site. Amazon did not give merchants a choice or a commission structure. The listings are generated without a fee arrangement. Amazon says it’s not charging merchants a commission during beta; the merchant is essentially getting incremental sales without paying Amazon. But Amazon gains value by keeping user engagement and capturing data.

How did merchants find out they were included?

Most found out the hard way: unexpected orders started coming in. Amazon did not notify them ahead of time or ask permission. As for compensation, merchants do receive payment for any orders Amazon places on their site (just like a regular customer sale, minus whatever payment processing fees they normally pay). However, Amazon does not take a marketplace commission on top – it’s not like a 15% fee as in a typical Amazon sale. Amazon’s “gain” is keeping the customer on its platform and potentially earning their loyalty (and capturing data). The merchant gets the revenue from the product sale, but they didn’t explicitly agree to Amazon being a sales channel.

Is it legal for Amazon to list and sell products from other websites without permission?

Legality in this context is a gray area because Amazon isn’t stealing the products; it’s acting as a customer would. If you have a public online store, anyone (including a bot) can technically place orders. Amazon is leveraging that, along with publicly available information. There’s no specific law against listing information found on the web, especially if it’s factual like a product name and price. However, there could be intellectual property questions (using product images or descriptions without permission) and contractual issues (for example, if a brand’s terms of service prohibit automated scraping or resale, Amazon could be in breach of those terms). No major legal action has been taken publicly as of now, but many affected brands feel it’s unethical. It’s possible this area will attract regulatory scrutiny if it grows, since it touches on competition and consumer transparency as well.

Why are merchants so upset if they’re making sales through Amazon’s Buy for Me?

For many merchants, it’s not just about the sale – it’s about control and consent. They’re upset because: (1) They didn’t agree to have their brand represented on Amazon, yet it was. (2) Some deliberately stay off Amazon to curate their brand image or pricing, and this undercut that choice. (3) issues like wrong info or out-of-stock orders made their business look unreliable, and they had to deal with angry customers. (4) They lose the direct relationship with customers (Amazon keeps the customer’s info and engagement). So even if a few extra sales come in, the cost to their brand reputation or long-term customer strategy can be negative. It’s analogous to finding your products being sold in a store you never approved – even if money comes in, you’re concerned about how they’re being sold and presented.

How do merchants remove their products from Buy for Me?

Amazon has provided an opt-out, though it’s not widely advertised. A merchant can contact Amazon (for example, via a specific email like branddirect@amazon.com) to request their site or products be removed from these programs. Merchants have reported that Amazon did comply and took their listings down within a few days of opting out. In the meantime, some have also taken measures like canceling any orders that come through Amazon’s bot (so the customer doesn’t get the item via Amazon) while they sort out the removal. Unfortunately, the onus is on each merchant to opt out if they don’t want to participate – it was an opt-out program by default.

How do I gift an item on Amazon?

To gift an item on Amazon, add it to your cart, proceed to checkout, and select “This order contains a gift.” Enter your friend’s address as the shipping destination. Selecting “This order contains a gift” hides prices on the physical packing slip. You can add a free gift message and, if available, select paid gift wrapping as options. Note that some third-party sellers may not offer gift wrapping or messaging, and a notice will appear during checkout if these options are unavailable. The gifting feature requires an Amazon Prime membership for shipments within the continental U.S. If you do not know your friend’s address, the Amazon app allows you to enter their email or phone number to send the gift. Recipients can exchange the gift for an Amazon Gift Card without notifying the sender.

What is the Universal Commerce Protocol (UCP) mentioned in this context?

The Universal Commerce Protocol is an open standard developed by companies like Google and Shopify. It’s basically a structured way for merchants to allow AI agents to transact on their sites. Through UCP, a merchant publishes how an AI can discover products, check inventory, and complete a checkout, all with explicit permission and standard rules. Think of it as a common language that could let, say, Google’s shopping assistant buy an item from a boutique’s website seamlessly, with the boutique’s blessing. UCP is meant to ensure any AI shopping action is consensual and that the merchant stays in control of product info and checkout conditions. It’s the polar opposite approach to what Amazon did with Buy for Me. With UCP, the merchant opts in and actively participates; Amazon’s approach was opt-out and done without initial consent.

Would UCP prevent something like Amazon’s Buy for Me?

Not automatically. UCP isn’t a law or a physical barrier – it’s a voluntary standard. If a platform like Amazon chooses not to honor it (or not to participate in it), they can still do their own thing like scraping sites or acting as an agent without permission. UCP works if all parties agree to use it. In the current scenario, Amazon has not joined UCP, so it’s essentially doing an end-run around these emerging standards. However, if UCP gains widespread adoption and merchants signal their preferences through it, one could imagine future where ignoring it might draw more backlash or even be addressed by regulators or industry norms. Today, UCP doesn’t “stop” Amazon; it simply offers a better path that we hope platforms will follow. It’s like the difference between an agreed-upon traffic law versus one driver deciding to go off-road – the law guides cooperative drivers, but it doesn’t physically stop a rogue actor.

How does Amazon’s AI assistant (Rufus) factor into Buy for Me?

Rufus is Amazon’s AI shopping assistant built into their app and website. It’s designed to help customers find products and answer questions. As part of its capabilities, Rufus can utilize features like Shop Direct and Buy for Me. For example, if you asked Rufus, “I need a red leather wallet under $100,” and the best match isn’t sold on Amazon, Rufus could show a Buy for Me result from an external brand and even execute the purchase. The key thing to note is that Rufus, being an Amazon tool, is aligned with Amazon’s marketplace. It will try to keep you shopping within Amazon’s services (including these agentic purchases). Unlike a neutral AI that might truly search the whole web and respect each site’s preferences, Rufus will favor Amazon’s ecosystem. So in a way, Rufus + Buy for Me together illustrate Amazon’s closed approach to agentic commerce: their AI will push Amazon-controlled solutions (even if the product is technically from an outside store, the experience remains in Amazon’s app).

What does this mean for the future of online shopping?

It indicates that a major change is underway. We’re moving from just finding things with AI to actually buying via AI agents. In the near future, you might commonly use an AI assistant to handle shopping tasks – from researching to comparing to purchasing – across multiple stores. The big question is whose terms will that future run on. Amazon’s experiment suggests one future where big platforms do it all for you (with some heavy-handed tactics). The alternative being built by others is a more open network where your agent could shop anywhere with merchants’ cooperation. For consumers, AI-driven shopping could be incredibly convenient. You could say “buy me a refill of my favorite shampoo from the cheapest source” and your assistant handles it. But behind the scenes, whether that transaction respected the merchant’s rules, or whether it cut them out, depends on which approach wins out. What’s clear is that online retailers need to prepare for AI-driven transactions – ensuring data accuracy, deciding on participation in protocols like UCP, and thinking about how to maintain customer relationships in a world where the “point of sale” might be a conversation with an AI. The Buy for Me incident is a bit of a warning shot that these changes are no longer theoretical; they’re happening now, and businesses large and small will have to adapt.

• Category: Uncategorized
• Tag: AI in Ecommerce, Amazon, Strategy