Salesforce's Agentforce Commerce just went GA. Here's what it means if you're not on Salesforce.

What actually shipped

• Shopper Agent: carries one customer conversation from discovery through checkout and into post-purchase service, speaking in the brand's voice on the brand's own storefront rather than handing the shopper off to a generic assistant.
• Buyer Agent: aimed at B2B, meeting buyers inside WhatsApp and SMS instead of a portal login.
• Merchant Agent: lets a commerce team organise catalogue, adjust pricing and respond to demand signals in plain-language prompts instead of a rules engine.
• ChatGPT and Google Search / Gemini integration is described as "arriving in the coming weeks", not yet live at GA. That is the piece to watch: it is the step that turns this from a better storefront assistant into a distribution channel Salesforce merchants did not have to build themselves.

Why this matters if you run Shopify, BigCommerce, WooCommerce or Wix

The specific product is Salesforce's. The signal is everyone's. A platform vendor with enterprise sales leverage just told its largest customers that the shopping agent, not the chatbot widget, is the next default surface for their storefront, and backed it with a general-availability date instead of a roadmap slide. That is the same shift our own agentic commerce primer for merchants and the agentic commerce glossary entry have been describing since late 2025: AI agents increasingly compare, choose and complete purchases with less human clicking, and the merchants who show up in that flow are the ones whose product data and evidence (reviews, UGC, specs) an agent can actually parse and quote.

None of that requires Salesforce. The shopping agent pattern (an AI shopper that reads a catalogue, checks reviews, and calls tools to complete a purchase) is architecture, not a single vendor's SKU. Every merchant's real question is whether their own storefront gives any agent, Salesforce-native or otherwise, something citable to work with.

What to actually do about it this week

1. 1Make sure product pages carry real Product schema, not just visual price tags, so an agent (or Google) can parse price, availability and reviews without guessing.
2. 2Get an AEO brand report or run the AI visibility check to see whether ChatGPT, Perplexity and Gemini can currently describe your products accurately at all.
3. 3Publish real, specific customer evidence (reviews with detail, UGC with product tags) on the PDP itself. Agents quote specifics, not marketing copy; our reviews-as-evidence research covers why that content gets cited disproportionately.
4. 4If you have any interest in MCP or an agent-callable product feed, this is a reasonable moment to build it, before "does your storefront support agent checkout" becomes a question every platform vendor asks by default.