An AI storefront is a store an AI can shop

The confusion is real, so settle it first.

• A storefront agents can shop. Product data, policies, and pages structured so AI assistants can parse them and transact. Shopify's Agentic Storefronts, New Gen's Kepler, and Kinect's Agent-Ready Storefront live here.
• A conversational storefront. Chat as the primary shopping interface on your own site. Swap's Storefront product and iAdvize sell this, and it overlaps with the first meaning more every quarter.
• A store built by AI. Prompt-to-theme generators like PagePilot and Wix's builder. A different job entirely. If that is what you came for, the category runs from free tiers to about $299 a month, and the rest of this page is about the other two.

One more naming trap. Storefront, the company at storefront-ai.com, sells monitoring of how AI assistants describe your brand. Useful, and a different product again.

Adobe has tracked AI-referred retail traffic since 2024, and the 2026 curve is the argument. First-quarter traffic to US retail sites rose 393 percent year over year. July alone rose 62 percent, and AI-sourced visitors have converted better than everyone else for 11 consecutive months, about 60 percent better in the July read, with 59 percent more time on site and a third less bouncing.

The same analysis carries the uncomfortable half. Homepage visibility to shopping agents averages 61 percent, meaning nearly 40 percent of what a merchant publishes is not fully readable by the assistants sending that converting traffic. Product pages read at 66 percent. FAQ and returns pages, the boring ones, read above 80.

Morgan Stanley's research desk sizes agentic shopping at $190–$385 billion in US ecommerce by 2030. Whether that lands high or low, the direction settled hard enough that Shopify rebuilt its distribution stack around it.

Shopify announced Agentic Storefronts on December 10, 2025 and activated the system for eligible merchants on March 24, 2026. Products flow through Shopify Catalog, which infers categories, extracts attributes, and consolidates variants, then surface inside ChatGPT, Microsoft Copilot, and Google's AI Mode and Gemini through the Universal Commerce Protocol. Orders land back in the admin with AI-channel attribution, and Shopify's own framing is no separate integrations, no apps, and no transaction fees beyond standard processing.

Two edges of the rollout matter for planning. The Agentic plan lets brands not hosted on Shopify list catalogs in the same index. And the Storefront MCP server exposes search, cart, and checkout to developers, so anyone can build a shopping agent against a store directly.

What the rollout does not do is make your own storefront readable. Catalog syndication answers the assistant that asks the channel. The assistant that browses your actual site, reading your pages, your policies, your sizing tables, still meets a store built for human eyeballs. That gap is the 61 percent number, and closing it is a merchant-side job.

The field, from each vendor's own published materials.

Readable product data comes first. Attributes filled in, variants consolidated, fit and materials stated in text an agent can parse rather than baked into images, policies published where a machine can find them. The stores scoring above 80 on Adobe's visibility read treat their pages as data, and that catalog work is the same work that makes enrichment pay off across every channel at once.

Then the selling layer. An agent-readable store answers the buyer's assistant correctly. A brand agent goes further and sells, in the brand's voice, to the human shopper and the AI one. That lineup is what Kinect runs as one intelligence trained on your catalog, policies, and conversations.

On outcomes, Kinect publishes its method rather than a promise. Brands with Kinect see 3–6% more revenue, measured against their own baselines, and shoppers who engage convert at 6–21%, an engaged-cohort read rather than a causal claim. The method is documented at how we measure.