By Sophie Lindqvist

Muse passed 2.5 million downloads in its first two weeks, per Sensor Tower data reported by CBS News, and briefly knocked ChatGPT off the top of Apple's App Store. That is the number marketing teams should sit with, because consumer AI agents now sit between brands and the people brands thought they owned.

Meta's Muse isn't the first consumer agent, and it looks a lot like its open-source predecessor, OpenClaw — Meta has admitted the resemblance isn't a coincidence. The agent can compare products, evaluate policies, and buy things without the customer ever visiting your website. Brands now face a choice: compete when consumer agents mediate shopping, or lose those customers to competitors that do.

How did retailers respond?

Badly, then inconsistently. On September 20, 2026, Amazon started showing Muse users a pop-up warning that an "unauthorized AI agent" violates its Conditions of Use. Two days later, PayPal announced checkout through Muse at its merchants worldwide. Shopify had enabled Shop Pay checkout for Muse across all its stores the day before that. Three major players, two opposite decisions, inside a week.

Muse has company. OpenAI launched Dots in late September, an always-on assistant inside ChatGPT that runs background workflows across connected apps; it requires a paid Pro or Business Premium plan and isn't available in the EEA, U.K., or Switzerland. Google has Gemini 4 Argon in limited preview through its Fairwind Program, with no consumer release date. Apple's agentic Siri works across on-device apps, and Anthropic offers desktop control via Claude's Cowork app on Pro and Max plans. Each sets its own rules for how its agent identifies itself. Your access policy has to account for all of them. That's the awkward part.

What does the law say?

Amazon's objections are specific. The company told GeekWire that Meta never notified it Muse would access the store, that the agent doesn't identify itself while browsing, and that it appears to capture customer credentials. Amazon can afford that stance — it generated more than $68 billion in ad revenue last year and runs its own shopping agent, Buy for Me, which identifies itself and lets brands opt out. Most brands have neither.

The more common case looks like Target's, where a CNN reporter testing Muse could browse but had to enter personal details by hand because the retailer blocked autonomous clicks at checkout. At many companies, that decision starts as a bot-management rule from the security team; marketing learns about it when conversion drops.

In August, the Ninth Circuit ruled in Amazon's case against Perplexity that the "user" is the one accessing a site under federal anti-hacking law, with the AI company acting on the user's behalf. That limits anti-hacking claims against agents, though terms of service remain a lever, as Amazon's warning shows. The takeaway: blocking agents is legally harder than it looks.

Is there a middle path?

Yes, and the networks built it. Visa's Trusted Agent Protocol, developed with Cloudflare, lets merchants verify signed agents and separate them from malicious bots. Mastercard requires agents to register before transacting, and every European issuer is now enabled for Agent Pay. A brand can admit verified agents at checkout while still blocking scrapers. Amazon argued third-party agents "should operate openly" — most brands would agree, and now they can enforce that standard without slamming the door.

What happens to your data and measurement?

Meta designed Muse to remember what matters to the consumer, including details mentioned once, and keeps a full audit trail for the user; Meta says it doesn't share that data with its ad systems. The preference data brands collected for a decade through on-site behavior now accumulates somewhere the brand can't see.

Measurement breaks first. Per The Keyword's reporting on Shopify's Meta channel terms, Google Analytics and custom pixels don't fire in Meta's direct checkout — Shopify passes only server-to-server events for checkout started and completed. CDPs built around browsing sessions capture nothing of an agent checkout. Checkout.com's 2026 research found 72% of U.K. and U.S. merchants agree consumers will adopt agent-led shopping faster than merchants can accommodate. Deloitte's survey of 13,500 European consumers found 31% open to leaving the final purchase to AI and 47% unwilling. Brands that treat today's reluctance as permanent will repeat the mobile cycle.

How do you make your brand agent-friendly?

Agents hold your policies to their own standard. Muse checks out with Stripe's Link, the first agent covered by Link's purchase protections — damaged items, price drops, no-fee returns. If your return policy runs 30 days on the web, 14 in the app, and final sale on marketplaces, an agent will notice in seconds and rank a consistent competitor first. You'll never see the abandoned cart. Call it coherence at velocity: agents need specific, consistent, quantifiable terms on every transaction.

Accenture's "Talk to My AI Agent" research found 56% of consumers would tell an agent which brands to consider, while 37% of behaviorally loyal consumers would accept a switch if the alternative fit better. Preference survives when the agent can verify it — consistent terms, accurate inventory, loyalty benefits that apply without a login.

Practical first steps:

• Audit your defaults: Shopify shares products with Meta via Shopify Catalog by default, and direct checkout is on for eligible stores.
• Decide what verified agents may do at browsing, cart, and checkout stages, using Trusted Agent Protocol or Agent Pay signals instead of blanket blocks.
• Publish return, shipping, and warranty terms as machine-readable structured data, identical across channels.
• Make loyalty portable — let an agent pass a loyalty ID and apply benefits at checkout.
• Rebuild measurement for server-side events before you explain a surprise Q4 gap.

PayPal supports WebMCP, an emerging standard for agent commerce through existing storefronts. "We're not ready" no longer buys you a planning cycle. Amazon can hide behind a wall and its own agent; everyone else competes on what the agent can read. Simple as that.