Skip to content
New

Shopify launched Agentic Storefronts. We make AI agents recommend you - not just list you.

See the Shopify integration
Tru Commerce
PricingStart free →

USE CASE · CLAUDE

Claude buys through tools. If your catalog is not a tool, it cannot be bought.

Claude reaches commerce through MCP - connecting to catalogs exposed as structured, callable tools. Tru Commerce deploys your catalog as an MCP server so Claude can find your products, compare them, and complete the order.

measured first-party·your brand, your data·no redirect

Claude's commerce path runs through the Model Context Protocol rather than a crawled index. Brands with an MCP endpoint are reachable; brands without one are invisible to it regardless of how strong their SEO is.

PROOF

7×

MyMuse's real AI-driven revenue versus what GA4 reported - ₹81.2K/mo actual against ₹11.5K/mo attributed (Tru Commerce case study)

On Claude

Discover. Recommend. Transact.

01 · Discover

Expose your catalog over MCP so Claude can query it directly, with structured product truth.

02 · Recommend

Shape the tool responses - attributes, availability, comparisons - so Claude recommends accurately.

03 · Transact

Complete the order through the same connection, with you as merchant of record.

What powers it

The Tru Commerce products behind Claude

How Claude works

What decides whether it recommends you

Claude for Commerce ships the agent, not the order

Anthropic's 2 September 2026 blueprint open-sources a consumer shopping agent and merchant operations agents with eight reference examples (Retailgentic). It builds the conversation and the cart from your catalog; checkout completion, binding price and stock, fraud rules and coverage of the other surfaces are still yours to add.

MCP is the connection, not a crawl

Claude reaches external commerce through tools it can call, defined by the Model Context Protocol. This is a deliberate integration rather than something you earn through content.

Tool design shapes the recommendation

What your MCP server returns - which attributes, which comparisons, what availability - directly determines how well Claude can recommend your product against alternatives.

The buyer skews considered

Claude is heavily used for research and comparison work. Answers reward accurate specification and honest trade-offs over marketing language.

One deployment, many clients

An MCP server is not Claude-specific. The same endpoint serves any MCP-speaking agent, which makes it the highest-leverage single integration available today.

What changed on Claude

The last 60 days, with sources

Updated 2026-10-13. Each item carries its own source and date; items marked observed are walkthroughs someone ran on the surface, not company disclosures.

2026-09-02

Anthropic released Claude for Commerce

An Apache 2.0 reference blueprint, not a product: a shopping agent a business embeds in its own app or site and a merchant agent its staff use, with four fictional verticals (retail, travel, telecom, ticketing) and a Claude Code plugin. The shopping agent searches the merchant's own catalog and ends at a checkout URL the model never sees; payments, ranking, authorization and fraud rules are the deployment's. Shopify published its own implementation over UCP the same day. Anthropic's "carts up to 35% larger" and "60% more likely to complete a purchase" describe unnamed customers.

Source: Anthropic, claude.com, 2 Sep 2026; github.com/anthropics/commerce-agents

2026-08-18

Retailers whose largest AI referrer is Claude rose from one to 15 in a quarter

ReFiBuy's Q2 AI1000, co-developed with Digital Commerce 360, counts Claude-led retailers at 15, up from one in Q1. Over the same quarter ChatGPT-led retailers fell from 844 to 722 and Gemini-led doubled from 16 to 32.

Source: ReFiBuy AI1000 Q2 2026, 18 Aug 2026

2026-09-11 · observed

What the reference agent actually sees of your product

After reading the repository, ReFiBuy's Vedant Gupta followed one shopper sentence through the consumer agent: "What the agent gets is a short row of fields from the catalog", and the product page itself "does not really come into it". The tool response is the product, as far as the agent is concerned.

Source: Vedant Gupta via Scot Wingo, LinkedIn, 11 Sep 2026

2026-09-17

Merchant tooling is already wiring into the Claude Commerce Agent

ReFiBuy announced a bi-directional connection built for Anthropic's 2 September blueprint: enriched product data flows into a merchant's Claude commerce agent and anonymised shopper context flows back. Product data in; the order path is still the merchant's.

Source: ReFiBuy, 17 Sep 2026

Case study

GA4 showed ₹11.5K/mo. Real AI revenue: ₹81.2K/mo. 7× hidden

FAQ

Common questions

Anthropic's open-source blueprint for shopping agents, released 2 September 2026: a consumer agent that recommends and builds a cart from a store's catalog, and merchant agents for store operations, with eight reference examples across retail, travel, telecom and entertainment. Shopify is a launch partner. It does not include checkout; the order still runs on the merchant's rails (Retailgentic, 2 Sep 2026).

← All AI surfaces

Find out where you stand on Claude.

Book a demo or talk to founders