Quick Summary
- •ESW launched Agentic Commerce with Microsoft Copilot on July 21, 2026
- •Brand catalogs become discoverable and transactable inside AI shopping experiences
- •ESW cites McKinsey $3-5T by 2030 agentic commerce opportunity and Gartner -25% traditional search
- •Action: assume agent-facing DTC checkout is a 12-month decision; make Amazon data agent-readable today via MCP
Nova surfaces every Amazon fee, refund, and margin shift in your live P&L, across 21 marketplaces. See it in your data
What ESW shipped on July 21
ESW, the cross-border ecommerce platform behind storefronts for a long list of global brands, launched ESW Agentic Commerce on July 21, 2026. The solution lets brand catalogs be discovered and transacted inside AI shopping experiences without touching the underlying ecommerce stack. Microsoft Copilot is the first integration, with more agents planned in the coming months (ESW, July 21, 2026).
The framing in the press release is the interesting bit. ESW cites McKinsey putting agentic commerce at a $3 to $5 trillion global opportunity by 2030 and Gartner projecting a 25% decline in traditional search as shoppers move to AI assistants (PR Newswire, July 21, 2026). Whether those specific numbers land is beside the point. The direction of travel is what matters, and every major ecommerce infrastructure vendor is building an agent-facing checkout layer this year.
How this fits with the wider agentic commerce push
Visa and OpenAI announced agentic payments earlier in June. Stripe published Sessions 2026 material on Google's AI commerce token billing. Shopify's Spring '26 catalog work made storefronts more agent-legible. Amazon's own AgentCore stack is already open to external services. ESW's launch is another vote for the same thesis: the checkout inside the AI conversation is a real surface, and brands that ship to it early will earn discovery ahead of brands that wait for a mature standard.
What Amazon sellers should do about it
- Assume agent-facing checkout is a 12-month decision, not a 3-year one. If you run meaningful DTC volume alongside Amazon, the question is when you expose your catalog to Copilot, ChatGPT, Gemini and Claude, not whether.
- Make your Amazon data readable by an agent today. A shopper who asks Claude "how did my Amazon business do this week" should get a real answer grounded in the actual Seller Central data, not a plausible-sounding hallucination.
- Separate the two use cases. Agent-facing storefront (Copilot buying) is a DTC front-end problem. Agent-facing analytics (Claude asking about your business) is a Seller Central data-access problem. Different vendors, different budgets, different owners internally.
- Track referrals from AI assistants as a distinct traffic source in whatever analytics you already run. The volume is small this quarter and worth measuring so the trend line is visible when it flips.
How Nova helps
- Nova MCP for Claude - connect your Amazon Seller Central data to Claude via the Model Context Protocol so agent-driven analytics questions get real answers, not guesses. Same architecture works with other MCP-compatible agents as they roll out.
Frequently Asked Questions
Common questions about this topic
Verified Sources
- ESW: ESW launches agentic commerce with Microsoft Copilot (July 21, 2026)
- PR Newswire: ESW launches agentic commerce with Microsoft Copilot
All information verified from official Amazon sources and trusted industry analysts as of publication date.
Gemini
ChatGPT