Vercel Shop makes AI agents first-class shoppers with agent-readable Shopify storefronts

Vercel has made Vercel Shop generally available: an open-source Next.js storefront template for Shopify designed around the idea that AI agents should be treated as first-class shoppers, not as bots forced to reverse-engineer pages built for people. According to the Vercel blog, the release was announced alongside a conversation between Shopify CEO Tobi Lütke and Vercel CEO Guillermo Rauch about the future of agent commerce at Vercel Ship SF.

Machine-readable product data on the same URL

According to Vercel, product pages built with the template expose machine-readable content covering price, availability and options, served from the same URLs that render visual pages for humans. An agent can read the product information directly instead of interpreting a page's layout. The template also embeds structured data in pages and sitemaps so that search and answer engines can discover and understand a store's catalog.

A rebuilt SDK and a shared commerce protocol

Vercel Shop is built on Shopify's Hydrogen SDK, which gives development teams access to a store's product catalog, cart and order management, payments and checkout. The two companies announced earlier this year that Hydrogen would be rebuilt to work across JavaScript frameworks, and the Vercel blog says the rebuilt SDK is now generally available.

Shopify also co-developed the Universal Commerce Protocol (UCP), described as a standard way for agents to search a store's catalog, read product details, check availability, build a cart and receive a checkout link. Vercel Shop makes these capabilities discoverable on the merchant's own domain, so agents can use MCP to search products and assemble carts without navigating the site itself.

One cart for browsing and chat

The template ships with a shopping assistant powered by eve, living in the same codebase as the storefront. Vercel says teams can customize both the store's design and how the assistant helps shoppers, and because the assistant and storefront share a single cart, items persist as a shopper moves between browsing product pages and asking the assistant to find things.

The template also includes skills that guide coding agents through setup, customization and template updates. To demonstrate the stack, Vercel launched its own merch store at vercel.store with limited-edition drops, including a shopping agent that helps shoppers find products, add them to a cart and proceed to checkout.

Early results

Shopify, Next.js and Vercel already power storefronts for brands including AG1, Bombas and PAIGE, according to the post. When PAIGE re-architected its ecommerce platform, it reported a 22% increase in revenue and a 76% increase in conversion over Black Friday and Cyber Monday, with no downtime.

Why it matters

Most commerce infrastructure on the web still assumes a human sitting at a browser. As AI agents increasingly browse and buy on users' behalf, they work around that assumption by parsing pages meant for people or relying on a small set of platform APIs. Vercel and Shopify are proposing the opposite arrangement: give agents structured product data, a standard protocol exposed at the merchant's own domain, and MCP-based tools to search and build carts directly. For merchants, that is a way to stay in the loop as agent-mediated shopping grows rather than being bypassed by whatever storefront an agent's operator picks. Because Vercel Shop is open source and sits on a widely used stack, it could become a practical template for what an agent-ready storefront means. The shared-cart design also signals the near-term model: agents assisting the shopping experience rather than replacing it.