Anthropic, the company behind Claude, has stepped directly into the commerce stack. On September 2, it released an open-source blueprint that lets retailers, travel companies, telecoms, and ticketing platforms build their own AI shopping and merchant agents on top of Claude, with Visa, Mastercard and Accenture signed on as launch partners. For an industry that has spent the last two years asking whether AI would just answer questions or actually transact, this is the clearest signal yet that the answer is the latter.

What Actually Launched

Anthropic is calling it a blueprint rather than a finished product, and that distinction matters. Released under an Apache 2.0 licence, the package gives engineering teams two pre-built reference designs, plus a Claude Code plugin, so a commerce agent can reportedly go from concept to running in days rather than months.

It is not a plug-and-play checkout system. Businesses deploy it through the Claude API, Amazon Bedrock, Microsoft Foundry, or Google Vertex, then connect it to their own catalog, inventory, and pricing data.

• Shopping Agents

Shopping agents are built for the customer-facing side. They search a merchant's catalog, compare products, remember a shopper's preferences across sessions, assemble a cart, and answer questions along the way.

• Merchant Agents

Merchant agents are built for the operational side. They recommend pricing and promotions, track inventory, and analyse sales performance, though any proactive change still requires human sign-off before it goes live, a guardrail Anthropic built in deliberately to avoid manipulative upselling or prices drifting from real catalog data.

Shopify has already published a public code repository showing the agents wired into a live storefront. Priceline is running one too. Beyond the four headline partners, Anthropic also lists Intuit, Klaviyo, Wix, Square, and Zomato among businesses already building on Claude for commerce use cases.

The Piece That's Missing on Purpose

Here's the detail that got buried in some of the early coverage: the blueprint ships with no payment protocol, no checkout flow, and no advertising layer. Anthropic built the reasoning engine, the part that understands what a shopper wants and assembles a cart, and left the actual movement of money to the merchant's existing checkout and to partners like Visa and Mastercard.

That's precisely why the payments giants are in the room. Their job is to answer the harder questions agentic commerce raises: how does a merchant verify that an AI agent is authorised to spend on a customer's behalf, what happens when a transaction limit is set, who's liable if an agent buys the wrong item, and how does a dispute get resolved when there was no human in the loop at the moment of purchase.

Visa's Jack Forestell framed the stakes simply, saying AI will reshape commerce but that "trust must remain at the centre" of every transaction. In other words: the interface may increasingly belong to AI, but the accountability framework around the transaction still has to be decided in the payments layer.

Do the Numbers Hold Up?

Anthropic has shared early figures from retailers already running shopping agents on Claude, carts reportedly up to 35% larger, and shoppers as much as 60% more likely to complete a purchase. Worth flagging plainly: those are Anthropic's own reported figures, not an independently audited benchmark, so treat them as a promising early signal rather than an industry-wide guarantee.

Separately, Accenture research cited alongside the launch found that a large majority of consumers are now open to collaborating with an AI agent while shopping, and a notable share say they'd trust a personal AI agent with a purchase decision more than they'd trust people close to them. Whether or not that finding fully generalises, it points to a real shift in consumer comfort with AI making purchasing calls, which is exactly the shift retailers now need a strategy for.

What This Means If You're Building or Marketing a Brand Right Now

• The Storefront Is No Longer the Only Entry Point

If a meaningful share of shoppers start discovering and comparing products through an AI agent instead of a homepage, your product data, pricing clarity, and catalog structure become part of your UX, not just your backend.

• Trust Signals Move Upstream

Agentic commerce doesn't remove the need for a brand to feel secure and credible; it just relocates where that impression gets formed, from a checkout page to an AI-mediated conversation the brand doesn't fully control.

• This Is a Payments-and-Product Problem, Not Just a Marketing One

Getting agent-ready means catalog hygiene, identity and transaction guardrails, and design decisions about what the agent is and isn't allowed to do, long before a single ad campaign gets written.

Why Agentic Commerce Matters Now

Agentic commerce is still early, and a blueprint isn't the same as a finished category. But when Visa, Mastercard, Accenture, and Shopify all show up to the same launch on the same day, it's a fair signal that this is where a meaningful part of the buying journey is heading next.

How Boğa Helps Brands Adapt

At Boğa, we help fintech, retail, and consumer brands adapt their digital experiences to emerging technologies, from UX strategy and product design to brand and digital growth. As AI continues to reshape how people discover, evaluate, and interact with brands, businesses need digital experiences that remain clear, trustworthy, and relevant.

Want to explore what these changes could mean for your brand? Get in touch with us at uxboga.com.