Agentic Commerce

Definition

Agentic commerce describes a new model of online shopping where AI agents act on behalf of consumers - discovering products, comparing options, and in some cases completing purchases - with minimal human intervention. Instead of a shopper browsing websites, clicking through categories, and manually comparing prices, an AI agent handles these tasks through direct access to product data and commerce APIs.

The term distinguishes this emerging paradigm from earlier forms of AI in ecommerce (like product recommendations or chatbots) by emphasizing agent autonomy. In agentic commerce, the AI does not just suggest - it acts. It searches across stores, filters results, checks availability, and can initiate checkout. The human’s role shifts from navigator to decision-maker.

Agentic commerce has accelerated rapidly since late 2024, driven by major technology companies building shopping capabilities into their AI assistants. OpenAI integrated shopping into ChatGPT. Perplexity launched a shopping agent. Microsoft built purchasing into Copilot. Amazon introduced “Buy for Me.” Google is building commerce features into Gemini. Each represents a different vision of how AI agents should handle shopping, but they all point to the same structural shift.

Why It Matters

Agentic commerce matters for merchants because it introduces a new discovery channel that operates by fundamentally different rules than traditional search or social media.

Discovery changes. In traditional ecommerce, merchants optimize for Google’s algorithm, bid on keywords, and build social media presence. In agentic commerce, the AI agent decides which products to surface based on structured data quality, relevance to the user’s query, and the agent’s ability to access the store’s catalog. SEO still matters, but the game is expanding.

The intermediary shifts. Google has been the primary intermediary between merchants and shoppers for two decades. Agentic commerce introduces new intermediaries - ChatGPT, Claude, Perplexity, Copilot - each with different data access, different ranking logic, and different commercial models. Merchants now need to think about visibility across multiple AI platforms, not just search engines.

Product data becomes the product. In agentic commerce, an AI agent never sees your beautifully designed website. It sees your product data - titles, descriptions, attributes, pricing, availability. The quality of that data determines whether your products get recommended. A store with thin product descriptions and missing attributes is invisible to AI agents, regardless of how well its website converts human visitors.

European merchants face a timing question. Most agentic commerce protocols and AI shopping features launched in the US first. European merchants have a window to prepare - ensuring their product data is AI-ready before these channels fully expand into European markets.

How It Works

Agentic commerce operates through a stack of technologies and protocols:

AI assistants (ChatGPT, Claude, Perplexity, Gemini) serve as the consumer-facing layer. Users express shopping intent in natural language: “Find me a waterproof hiking jacket under 200 euros, available in medium.”

Protocols (MCP, ACP, UCP) define how AI agents connect to stores and query product data. These protocols give agents structured access to product catalogs, pricing, and availability - far more reliable than web scraping.

Structured data (Schema.org, JSON-LD) on product pages provides a fallback discovery mechanism. Even without a direct protocol connection, AI agents that browse the web can extract structured product data from properly marked-up pages.

Product data quality is the merchant’s primary lever. The AI agent’s ability to match a product to a user’s query depends on how well the product is described. Detailed titles, comprehensive descriptions, complete attributes (size, color, material), accurate pricing, and real-time availability all feed into the agent’s decision-making.

The transaction flow varies by platform. Some AI agents generate checkout links that redirect to the merchant’s site. Others are building more integrated purchase flows. The common thread is that the AI agent handles discovery and evaluation, while the final purchase decision and payment remain with the human user.

Related Terms

• Model Context Protocol (MCP) - The foundational protocol enabling AI agents to access store data
• Agentic Commerce Protocol (ACP) - Shopify’s commerce-specific protocol for AI agents
• Generative Engine Optimization (GEO) - Optimization practices for AI-driven discovery
• AI Readiness - A store’s preparedness for AI agent interactions
• Universal Context Protocol (UCP) - Platform-agnostic protocol for cross-platform AI commerce