Consumer AI agents are no longer a pilot project. Meta’s Muse passed 2.5 million downloads in its first two weeks and briefly moved past ChatGPT at the top of Apple’s App Store, according to Sensor Tower data reported by CBS News. The speed of adoption is forcing a decision many brands hoped to postpone.
Retailers are already split
On Sept. 20, 2026, Amazon began warning people using Meta’s Muse to shop its store that access by an “unauthorized AI agent” violates Amazon’s Conditions of Use. The next day, Shopify enabled Shop Pay checkout for Muse across all of its stores. The day after, PayPal announced customers could shop and check out through Muse at PayPal merchants worldwide. Three major platforms made two different calls within days.
The agent field has only widened. OpenAI’s Dots, Google’s Gemini 4 Argon preview, Apple’s agentic Siri and Anthropic’s Claude Cowork are all moving into consumer tasks. Meta and OpenAI run a user’s agent in a dedicated cloud environment, so the browsing, comparison and checkout happen on infrastructure the brand never touches.
Why access is not just an Amazon problem
Amazon can afford a hard line. It generated more than $68 billion in ad revenue last year from people browsing its pages and runs its own shopping agent, Buy for Me. Most brands have neither asset. A blanket block keeps some manual shoppers on your site, but customers who continue with their agent take the transaction — and everything the agent knows about them — to a retailer that lets it in.
The legal backdrop shifted in August, when the Ninth Circuit ruled in Amazon’s case against Perplexity that the “user” is the person accessing a site under federal anti-hacking law and the AI company acts on that user’s behalf. Terms of service remain an option, but a middle path is emerging: Visa’s Trusted Agent Protocol and Mastercard’s Agent Pay let merchants verify signed agents and separate them from malicious bots without blocking all agent access.
What brands lose when the agent holds context
Meta designed Muse to remember consumer preferences, including details a person mentions only once. The first-party preference data brands once collected from on-site behaviour now accumulates in a private environment the brand cannot see. The impact shows up first in measurement: with Meta’s direct checkout, Google Analytics and custom pixels don’t fire, and Shopify passes only server-to-server events for checkout started and completed. The behavioural trail that fed CDPs never arrives.
Checkout.com’s 2026 research found 72% of merchants in the U.K. and U.S. agree consumers will adopt agent-led shopping faster than most merchants are prepared to accommodate.
Policy gaps now cost silent sales
Muse checks out with Stripe’s Link, and Meta says it is the first agent covered by Link’s purchase protections. That gives the consumer’s wallet a baseline the agent compares against a brand’s terms on every purchase. An apparel brand with 30-day web returns, 14-day app returns and final-sale marketplace terms may never see a bounce or abandoned cart; the agent simply ranks a competitor with one consistent policy first. Consistency across siloed channels becomes a “coherence at velocity” problem.
Brand preference is not dead. Accenture’s “Talk to My AI Agent” research found 56% of consumers would tell an agent which brands to consider, while 37% of behaviourally loyal consumers would accept an agent switching brands if the alternative fit better. Preference holds when an agent can verify concrete value: consistent terms, accurate inventory and loyalty benefits that apply without a human login.
Five first steps for marketing teams
- Audit your default settings — Shopify shares merchant products with Meta via Shopify Catalog by default, and direct checkout is enabled by default for eligible stores.
- Define agent access by funnel stage — decide separately what verified agents may do during browsing, cart creation and checkout, using Trusted Agent Protocol or Agent Pay signals.
- Make policies machine-readable — keep return, shipping and warranty terms identical across channels in structured data.
- Make loyalty portable — let an agent pass a loyalty ID and apply benefits at checkout.
- Rebuild measurement for server-side events — set up agent-channel reporting before the Q4 gap appears.
Brands whose information an agent can verify, and whose benefits are tangible enough for machines to understand, will compete for the next wave of customers. The default settings and policy gaps are the first thing to fix.
Source: MarTech
