AI Shopping Agents: What Amazon vs Meta Means for Marketers

Overview

AI shopping agents just had their busiest week. On November 6, 2026, Amazon let invited US shoppers text Alexa for Shopping to search, build a cart and check out, all inside a message thread. The same day, Meta added spaces to its Muse agent, one day after giving Muse a voice and a face.

The timing isn't an accident. Amazon has blocked Muse from its store since September 20, and both companies now want to be the agent people ask first. That's a big deal for marketers, because an agent that picks products for people also decides which brands they see.

This guide covers what each company launched and the numbers behind the race. It also looks at what the shift means for retail media and gives you a checklist for getting products ready before Black Friday on November 27.

Disclosure: Vanaxity builds AI agents for marketing, including tools that help brands show up in AI answers. We have a stake in this shift. This article was drafted with help from Claude. Its numbers come from company statements and press reports, and we label our own math.

Key takeaways

• Amazon now lets invited US customers shop by text with Alexa for Shopping, from search to checkout, with millions more due in the coming weeks.
• Meta's Muse added spaces, small tools people build by describing them, plus real-time voice and avatars in the US.
• Amazon has blocked Muse since September 20.
• Both sides report heavy use: 400 million Alexa for Shopping users in a year, and over 25% of Muse users shopping with it.
• Adobe forecasts $275.1 billion in US online holiday spending and a 130% jump in traffic from AI services.
• Marketers should make product data easy for machines to read, track AI referrals as a channel and test what agents say.

AI shopping agents: the race at a glance

Here's how the two companies moved over the last seven weeks, in date order.

What Amazon launched: Alexa for Shopping by text

Amazon's new feature works like texting a friend who knows the store. Invited customers accept the terms, link a phone number and get a contact called Alexa Shopping. From there, a text such as 'compare the top-rated robot vacuums' starts a search.

Amazon says the agent can handle these jobs inside the thread:

• Search for products and compare options side by side
• Add items to a cart and complete the purchase, with your approval
• Reorder things you've bought before
• Track prices and send an alert when one drops
• Share order updates and delivery notices

Chats carry over between text, the Amazon Shopping app, Amazon.com and Alexa devices. Access is invite-only for now, and Amazon says millions more US customers will get it in the coming weeks. Everyone else can join a waitlist.

One line in the launch matters for brands that don't sell on Amazon. Amazon says Alexa for Shopping can surface options from stores across the web through Shop Direct. So products from other sites can appear in its answers, though Amazon hasn't said how it picks them.

Why did Amazon block Meta's Muse?

Muse is Meta's personal agent. It can browse retail sites and buy things for people, which is exactly what Amazon objected to. On the night of September 20, Amazon began blocking Muse, and Muse users who tried to shop there got a pop-up about unauthorized agents.

Amazon gave three reasons, according to press reports:

• Meta never asked permission for Muse to use the store.
• Muse doesn't identify itself as an agent while it browses.
• The agent appears to capture and store customer passwords, a claim Meta disputes.

There's a business reason too, and it's about ads. When an outside agent shops for someone, that person may never see the sponsored products Amazon sells to brands. Forbes tied the block to that ad business. Amazon took a similar line with Perplexity's Comet browser agent, which it sued in 2025.

Muse still works elsewhere. Meta has signed up retailers that want it: Shopify lets Muse check out through Shop Pay, and Meta has named partners such as Walmart, Gap, Sephora, Wayfair and Best Buy.

What Meta added to Muse: spaces and voice

Meta Muse got two updates in two days. On November 5, it rolled out real-time voice and expressive avatars in the US, on mobile, the web and Mac. People can now talk to Muse out loud while an animated face reacts.

On November 6, Meta launched spaces. A space is a small tool that Muse builds from a plain description, like a holiday gift tracker, a trip planner or an order tracker for a small business. If you connect other apps, a space keeps itself up to date in the background.

For shopping, spaces could be the bigger deal. A gift tracker that pulls in wishlists and watches prices turns Muse from a one-off helper into a place people come back to all season. Meta hasn't said whether anything people put in a space, from wishlists to budgets, will shape the ads they see on Facebook and Instagram.

AI shopping agents by the numbers

Both companies shared numbers, and nearly all of them are self-reported. Here's the list.

Two of these deserve a closer look. Jassy's 50% figure compares people who chose to use the agent with those who didn't, so it may say more about who uses it than what it does. And Meta's 25% is a share of Muse's weekly users, a total it hasn't disclosed.

The holiday forecast for AI shopping traffic

Adobe forecasts that US shoppers will spend $275.1 billion online from November 1 to December 31, up 6.7% from 2025. Cyber Week alone should bring in $47.5 billion.

The faster number is AI traffic. Adobe expects clicks from AI chat and browser services to US retail sites to rise 130% from last year. The biggest jump, 159%, should come on Thanksgiving, ahead of 95% on Black Friday and 88% on Cyber Monday.

Keep the base in mind. AI traffic grew from a small start, so even a 130% rise leaves it a modest slice of visits. Adobe hasn't published its share of all retail traffic.

Two models for AI shopping agents: closed store vs open web

Amazon and Meta are building different things, and the gap shapes what brands should do.

If you sell on Amazon, Alexa for Shopping reads the same catalog your listings live in. If you sell direct, Muse and agents like it read your website and feeds. Most brands will need to show up well in both.

What AI shopping agents mean for retail media

Amazon's ad business brought in about $78 billion over the last four quarters, by PPC Land's count. In the third quarter, ad revenue rose 11% to $19.6 billion and missed Wall Street estimates for the first time in seven quarters.

Jassy said outside agents weren't the cause. He blamed Prime Day timing. Still, the risk is easy to see: sponsored listings work because people scroll past them, and an agent doesn't scroll.

Amazon's answer is to own the agent. If shoppers text Alexa instead of asking Muse, Amazon keeps control of which products appear, which brands get named and which ads, if any, ride along. It hasn't said how sponsored products would show up in text replies.

For ad buyers, two things are worth watching: whether Amazon adds paid placements to agent answers, and whether organic product data starts to count for more once agents do the browsing. This is a marketing view, not investment advice.

How AI shopping agents read your products

People skim photos and brand names. Agents read text and data. Neither Amazon nor Meta has published how its agent ranks products, so the safest bet is to make every fact an agent needs easy to find and hard to misread.

• Titles that say what the product is, not just the brand name
• Full attributes: size, material, color, fit and what's in the box
• Prices, stock and delivery times that match across your site, feeds and marketplaces
• Clear return and warranty terms on the product page
• Reviews and answered questions, since agents can quote them

On your own site, add Product structured data with offers, ratings and shipping details. Keep your Shopify catalog and Merchant Center feeds clean, because feeds are how partner agents like Muse find products. On Amazon, fill every listing field, including the ones shoppers rarely see.

A useful test is to read a product page as if you were a machine. Strip out the photos and ask whether the text alone answers what a shopper would ask: what is it, who is it for, what does it cost and when will it arrive?

A holiday checklist for AI shopping agents

Black Friday is November 27 and Cyber Monday is November 30. Here's what fits in the next three weeks.

• Week 1: audit your top 50 products for missing attributes, wrong prices and thin titles.
• Week 1: test the agents. Ask Alexa for Shopping, Muse and ChatGPT for products in your category, and note which brands they name.
• Week 2: fix the gaps, add structured data and check that feeds update at least daily.
• Week 2: set up an AI referral channel in your analytics. Otherwise these visits hide in referral or direct.
• Week 3: write gift-ready copy that says who it's for, what it costs and when it will arrive.
• Week 3: set an agent policy, then check that robots.txt and bot rules match it.

Start with the products that already sell best. Agents tend to repeat what the data supports, so a bestseller with clean data is the fastest win.

How to measure AI shopping agent traffic

AI referrals are easy to miss. Visits from chat apps can arrive as plain referrals, and some show up as direct traffic with no source at all.

Start with a custom channel group in GA4 that catches known AI sources, such as chatgpt.com, perplexity.ai, gemini.google.com and the Meta AI domains you find in your reports. Then compare that channel's revenue per visit and conversion rate with search and social.

Some sales won't show up as visits at all. When Muse checks out through Shop Pay, or Alexa buys on Amazon, the shopper may never load your page. Look for those orders in your Shopify and Amazon sales reports instead.

Last, test by hand once a week. Ask each agent the same five shopping questions and log which brands it names. It's crude, but it's the best view you'll get of answers you can't buy your way into yet.

Should you block or allow AI shopping agents?

Amazon chose to block. It can afford to, because it owns both the store and its own agent. Most brands can't, and blocking an agent can mean losing the sales it would have sent.

A middle path works for most shops:

• Allow agents that identify themselves and follow your rules
• Rate-limit or block unknown bots that hide what they are
• Prefer agents that pay through a checkout partner, such as Shop Pay, over ones that use a stored password
• Review the choice each quarter as the rules settle

For more on consent and payments, read our guide to agentic commerce.

Worked example: what a 130% rise means for one shop

This is our own math, with made-up numbers. Say a shop had 500,000 visits last November and December, and 5,000 of them, or 1%, came from AI services.

If that AI traffic grows 130%, as Adobe forecasts, the shop gets 11,500 AI visits this season. Run them through a 2% conversion rate and an $80 average order, and they bring in about $18,400 in sales, up from $8,000 last year.

Real money. But it's still a small slice of the season for most shops, and the bigger shift sits outside these numbers: orders an agent completes without a page view, and products it never mentions at all.

Risks and open questions for AI shopping agents

Plenty is still unproven.

• Trust: a Gartner analyst told Marketing Brew it won't take much buyer's remorse to sour people on agents.
• Access: Amazon's text feature is invite-only, so early results come from a small group.
• Self-reported data: the 400 million and 25% figures come from the companies, with few details.
• Data use: Meta hasn't said whether spaces feed its ads, or how long it keeps voice chats.
• Security: Amazon and Meta still disagree about how Muse handles passwords.
• Value: Jassy says outside agents are 'still not great,' while Meta points to $200 orders.

What to watch next

If you want help getting product data ready for AI agents, see what we build or browse more Vanaxity insights. For Meta's wider ad picture, read our take on Meta's Q3 2026 earnings, and for the paid side of AI search, see our guide to ChatGPT Ads bidding.

AI shopping agents FAQ