For most of the last two years, AI shopping meant asking a chatbot for product ideas and then going off to buy them yourself. That changed in the space of a few weeks this fall. Meta, OpenAI and xAI all put agents in consumers' hands that can browse stores, fill carts and, with the shopper's approval, check out.
Then Amazon blocked one of them. The fight that followed says a lot about where shopping is heading and what brands should do about it.
What is a consumer shopping agent?
A consumer shopping agent is AI software that shops on a person's behalf. Instead of answering a question and handing over links, it visits stores, compares options, builds a cart and completes the purchase, usually asking the shopper to approve the final step. Most run in the cloud on their own virtual computer and browser, so they can keep working while the shopper does something else.
Meta Muse, OpenAI Dots and Grok Bot
Meta Muse
Meta launched Muse on September 8, 2026 as a personal agent with its own cloud computer. It can choose products and complete checkout, and purchases require the user's approval. According to Meta, Muse never sees users' passwords or payment details, which are kept in separate secure storage. Muse passed 2.5 million downloads in its first two weeks, according to Sensor Tower data, and Meta has announced work with Shopify and PayPal so Muse can complete purchases with participating merchants.
OpenAI Dots
OpenAI introduced Dots at DevDay on September 29, 2026. Each dot is an always-on agent with its own cloud computer and browser, connections to thousands of apps and the ability to research in the background. Dots are not built only for shopping, but an agent that runs all day, learns preferences and can use a browser will inevitably do a lot of comparing, tracking and buying.
Grok Bot
xAI's Grok Bot entered beta in August 2026 with the ability to browse online stores and complete purchases. It pays with single-use virtual card numbers through Stripe, and each purchase needs the user's sign-off on their phone. Grok was already powering Go, the AI shopping assistant Gopuff launched in June, which builds a cart from a description like hosting a game-day party.
Why Amazon blocked Muse
On September 20, Amazon blocked Muse from its store after Meta declined a request to leave Amazon out. Amazon's stated position is that apps buying on a customer's behalf should operate openly and respect a store's choice about whether to take part. It cited security concerns and said Muse did not identify itself as an AI agent.
There is a commercial side too. Amazon earned around $68 billion from advertising last year, and that business depends on people browsing its pages and seeing sponsored products. An agent that picks products on its own skips the ads. Amazon has its own agents, including Rufus and Buy for Me, and has previously moved against ChatGPT's shopping research and Perplexity's Comet browser.
Not every retailer is taking the same line. Within days of the block, Walmart, Best Buy, Sephora and Wayfair were named as Muse partners. Two strategies are forming. Keep agents out and protect the store you control, or let them in on clear terms and compete for the business they bring.
How shopping agents will change shopping behaviour
- Fewer browsing sessions, more delegated tasks. Reorders, gift lists, price watching and comparing a dozen options are exactly the jobs people will hand to an agent first.
- Comparison becomes instant. An agent can check price, shipping, stock and return terms across many stores in seconds. Habit and convenience count for less, fit and value count for more.
- The shortlist moves upstream. Shoppers will often see three options an agent picked, not your homepage or your category page. If you are not on the shortlist, you are not considered.
- Paid placement loses some power. Sponsored listings and on-site merchandising matter less when the agent is not looking at the page. Product data, reviews and policies matter more.
- The final click still matters. Muse and Grok Bot both ask the shopper to approve purchases. A clear total, honest shipping times and a smooth checkout still close the sale.
- Trust decides adoption. Shoppers are more comfortable letting agents research than letting them spend. Which agents people trust with their money will shape which ones win.
Should brands block shopping agents or let them in?
Amazon can afford to block agents because it is the destination. Most brands are not. For a brand doing $10M to $350M, an agent that can find and buy your products is a new route to customers, and blocking it hands that customer to a competitor whose store said yes.
Letting agents in does not mean letting anything in. The sensible middle ground is to welcome agents that identify themselves and set terms for how they behave.
- Identity. Agents should prove who they are. Standards such as Web Bot Auth, Visa's Trusted Agent Protocol and Mastercard Agent Pay exist for exactly this.
- Scope. Agents should act within limits the shopper set, such as a budget or an approval step.
- Payment through your checkout. Orders should settle through your normal payment flow so you keep the transaction, the data and the fraud controls.
- Aftercare. Your return and support policies should cover orders an agent placed, because the customer will call you, not the agent.
We go deeper on verification, fraud and chargebacks in when an AI agent buys from your store.
How brands can adapt
- Make your product data agent-ready. Complete attributes, consistent variants, current price and stock and clear policies decide whether an agent shortlists you. Start with our product data checklist for AI shopping agents.
- Decide your agent policy on purpose. Review your bot protection and terms of service so verified agents are allowed and bad bots are not. Many brands are blocking good agents by accident with rules written years ago.
- Use the channels your platform offers. Shopify merchants can manage AI channels through Agentic Storefronts, and partnerships like Muse's with Shopify and PayPal are opening more routes. Our guide to UCP, ACP and MCP explains the standards underneath.
- Keep checkout clean. An agent sends the shopper a summary to approve. Surprise fees, slow delivery estimates or a clunky checkout lose the sale at the last step.
- Own the relationship after the order. An agent may make the first purchase, but your packaging, email, loyalty program and service experience decide whether the second order comes back to you directly.
- Measure agent traffic. Track visits and orders from AI agents and assistants separately so you know whether this channel is growing for your brand.
- Ask your providers the right questions. As Rick Watson pointed out in his Watson Weekly newsletter, if you are choosing an agentic payments provider, ask which outside agents that provider allows on its own store. Its incentives may not match yours.
Common questions
What is Meta Muse?
Muse is Meta's personal AI agent, launched in September 2026. It runs on its own cloud computer and can choose products and complete checkout on a shopper's behalf, with the shopper approving purchases.
Can AI agents buy from my Shopify store?
Yes, increasingly. Shopify's Agentic Storefronts let merchants manage how products appear in AI channels, and agents like Muse, Dots and Grok Bot can browse and check out on regular storefronts. Whether they succeed depends on your product data, your checkout and your bot settings.
Should we block AI shopping agents?
Usually not. Block agents that will not identify themselves or behave badly, and let verified agents through. Blanket blocking mostly sends shoppers to competitors.
Will shopping agents replace our website?
No, but they change its job. Your site becomes the place agents read from and shoppers confirm on, so data quality, policies and checkout matter more than homepage merchandising.
How do we know if shopping agents are visiting our store?
Check your analytics and bot management logs for agent traffic and verified agent signatures, and track referral traffic from AI assistants. If you cannot see it today, setting that up is the first step.
Where to start
Check whether your store is turning agents away, then fix the product data that decides whether they shortlist you. If you want help building an agent strategy for your brand, get in touch.
