Over the past year, we’ve seen a meaningful shift in how AI is applied in commerce — away from isolated experiments and toward agents that can reason, act and drive real business outcomes. But one thing has remained clear: Most enterprise stacks aren’t built to support agents in any meaningful, secure or scalable way. That’s where Commerce MCP, our latest innovation, comes in.
At Elevate 2025, we introduced our answer to that gap – the Commerce MCP, a new capability that extends the commercetools Platform to make enterprise systems usable by AI agents — securely, modularly and in real time. This is not another chatbot wrapper. Its infrastructure is designed to meet a new interface paradigm: One where autonomous agents interact with APIs, not humans with screens.
From APIs to agents: Making commerce systems legible to AI
Most enterprise systems today weren’t designed to be interpreted by autonomous agents. They were designed for deterministic flows and user-driven interactions, not autonomous orchestration. That’s where Modern Context Protocol (MCP) comes in: It provides a standardized contract between AI agents and the business logic, services and data they need to act.
Our Commerce MCP uses MCP to expose key commercetools APIs — including carts, catalogs, pricing, promotions, inventory and orders — in a format that’s consumable by AI agents and governed for enterprise use. Instead of brittle integrations, you get a structured, secure way for agents to interpret business logic and trigger real workflows.
Built for the architect, not just the use case
We designed this extensibility layer to build intelligent AI agents for teams who think in architecture: Abstraction boundaries, security posture and long-term extensibility. It reflects the same principles that define commercetools: API-first, stateless, modular and built for interoperability.
That means:
Composable orchestration – Agents can work across commerce, CRM, service and marketing tools without coupling or duplication.
Governed execution – You decide what tools are exposed, how they’re scoped and who gets to invoke them.
Future-proof infrastructure – As MCP and agent-native clients evolve, your systems are already structured to meet them where they are.
What you can build today
This isn’t speculative. Customers are already prototyping and shipping new agent-based capabilities:
Shopping agents that use real-time inventory and pricing to guide personalized discovery and checkout, without needing to hardcode flows.
Developer copilots that can generate, validate or simulate commerce logic for testing and deployment.
Customer service agents that can look up real-time order data, initiate returns or escalate issues based on live system state — not canned responses.
Because these agents are built on protocol-aligned APIs, they scale without the fragility of traditional automation.
Furthermore, to help enterprises accelerate intelligent automation across systems further, APIs and workflows, the AI Hub provides plug-and-play integrations to leading AI platforms, starting with Microsoft Copilot Shopping.
Why commercetools, and why now
commercetools has spent over a decade helping enterprises decouple their commerce architectures so they can adapt faster. The Commerce MCP is an extension of that philosophy. It turns the commercetools Platform into an agent-ready environment — no rewrites, no workarounds, no disruption.
As frameworks like Claude Desktop, Cursor and LangChain gain enterprise adoption, the question shifts from “what can agents do?” to “what can they do with my stack?” With the Commerce MCP, you already have an answer.
What’s next
We’re launching early access with a select group of customers and partners. If you’re exploring agent-based applications or actively planning for them, we’d love to collaborate.
This is the next era of commerce — one where intelligent agents operate across your architecture, not around it. The infrastructure is here. The standards are emerging. The opportunity is real.
Let’s get to work.
Be among the first to explore the Commerce MCP and AI Hub — now in early access. Get in touch to learn more and secure your access.