AI models become platforms for apps and services
AI models become platforms for apps and services, influencing ecommerce with chatbots and interfaces.
Artificial intelligence models are becoming platforms, attracting an ecosystem of applications, connectors, marketplaces, and payment systems that could influence ecommerce. ChatGPT, Claude, Gemini, and similar services started as chatbots or interfaces for interacting with large language models. Now, these models are becoming platforms through which outside companies can provide software, data, and services.
This shift resembles the rise of earlier technology platforms. A useful product attracts users. Those users attract developers and businesses. Eventually, infrastructure emerges to connect participants, govern access, move money, and take a cut of the activity.
Anthropic offered a recent example in September 2026 when it launched Claude Marketplace, giving customers access to over 2,000 plugins and connectors, along with agents, software, and service providers. Enterprise customers can use some of their committed Anthropic spending to purchase products through the marketplace.
The Model Context Protocol (MCP) is one reason this ecosystem can develop relatively quickly. Anthropic introduced the MCP for AI systems to connect with outside software and data. In December 2025, Anthropic donated MCP to the Agentic AI Foundation under the Linux Foundation.
Block and OpenAI joined as founding contributors, and Amazon Web Services, Google, Microsoft, and other technology companies participate in the foundation. The move made MCP vendor-neutral and reduced one obstacle to adoption.
AI Marketplaces Emerge
Claude Marketplace gives software companies a way to reach customers already using Anthropic. In return, those integrations make Claude useful for more tasks without requiring Anthropic to build every application or data connection itself. The model resembles app stores and cloud marketplaces, although the economics are not yet clear.
Stratechery analyst Ben Thompson argued in September that AI agents could become powerful aggregators because users may rely on a relatively small number of agents to interact with many outside services. If that happens, businesses could compete not only for customers but also for an agent’s recommendation.
Ecommerce provides a useful example. AI tools already help shoppers research products, compare alternatives, and decide what to buy. The next step is connecting those conversations to merchant systems.
Google’s Universal Commerce Protocol provides a standard for commerce interactions involving AI agents. A merchant can make products available to Google’s AI surfaces and either support checkout there or send the assembled cart to its own website. Stripe has taken a similar approach with its Agentic Commerce Suite, which lets merchants make product information available to multiple AI agents while continuing to use their existing fulfillment and post-purchase systems.
Payment Infrastructure Develops
Payment infrastructure is developing too. Agent transactions create familiar payment problems, including authorization, fraud, refunds, and liability, but with an additional participant acting between the shopper and merchant. Google, Stripe, Visa, Mastercard, and others are developing programs to define what an agent can buy and how to authorize a transaction. No single approach has won, but that fact may be relatively less important for merchants today than the fact that the pieces are appearing.
McKinsey & Company estimates that by 2030 AI agents could influence or drive $900 billion to $1 trillion in U.S. consumer retail sales and $3 trillion to $5 trillion worldwide. For now, AI shopping is primarily about discovery and referral, not autonomous purchasing.
AI platforms could become another layer between merchants and customers, similar to search engines, ecommerce marketplaces, and social networks. The difference, however, is that an AI agent could do more than send traffic. AI can, to a degree, understand the shopper’s preferences, compare products across merchants, assemble a purchase, and eventually initiate a transaction.
Ecommerce Impact Expected
Apps, connectors, marketplaces, commerce protocols, and payment systems are beginning to form around AI platforms. For ecommerce companies, that emerging ecosystem may matter long before autonomous agents routinely make purchases.
