ACP vs UCP: the two agentic commerce protocols merchants must understand

UCP is Google's bet on an open, interoperable commerce web where any AI agent can discover and transact with any merchant. Built with Shopify, Walmart, Target, and major payment networks, UCP uses a manifest-based architecture where merchants publish their capabilities (search, cart, checkout, tracking) via a standardized /.well-known/ucp endpoint.

The protocol's strength is breadth: one integration theoretically makes your catalog discoverable across Google AI Mode, Gemini, Microsoft Copilot, and future agents. Shopify merchants get UCP automatically; others need custom integration work. Implementation typically takes 2-4 weeks depending on your existing API infrastructure.

ACP was built by OpenAI and Stripe specifically for conversational commerce — the "chat-to-buy" experience where users discover, compare, and purchase products within a single AI conversation. ACP uses shared payment tokens and Stripe's infrastructure to enable instant checkout without exposing card data to the AI assistant.

ChatGPT processed over $2.1 billion in ACP transactions in 2025, though conversion rates averaged 2.3x lower than native merchant websites, leading some major retailers to build dedicated ChatGPT apps that redirect to their own checkout instead of using ACP's instant purchase flow.

MCP isn't technically a commerce protocol — it's a context-sharing standard that lets AI models connect to external tools and data sources in a governed way. Developed by Anthropic and now governed by the Linux Foundation, MCP can support commerce through APIs but requires merchants to build their own transactional layer.

Think of MCP as the data layer that UCP or ACP might use for catalog access. It's powerful for enterprise AI applications but doesn't solve the commerce-specific challenges of checkout, payment authorization, or order attribution. Our MCP guide covers non-commerce use cases in detail.

AP2 (Agent Payments Protocol) solves the trust problem in agent-mediated payments: how do you verify that an AI agent has legitimate authority to spend on a user's behalf? AP2 uses cryptographic mandates and digital credentials to establish payment intent, cart contents, and transaction authority.

AP2 is the payment component within UCP's broader architecture — it handles the "can this agent really buy this?" question while UCP orchestrates the shopping journey. For most merchants, AP2 compliance comes automatically through UCP implementation rather than as a separate integration project.

VIC is Visa's intelligent commerce platform that enables AI agents to make purchase recommendations based on transaction history, merchant relationships, and spending patterns. Launched in April 2025, VIC focuses on the discovery layer — helping agents suggest products based on verified purchase intelligence rather than just catalog data.

VIC is in pilot with select merchants in Singapore and the UK. It's more of a recommendation engine that integrates with commerce protocols rather than a protocol itself. Most merchants will encounter VIC as an enhancement to their existing UCP or ACP implementations rather than as a standalone integration.