Agentic Commerce Glossary (A–Z)
Confused by a term? Our glossary defines key concepts and acronyms in AI-driven commerce – from AEO to UCP – with clear, concise explanations and references.
.
.well-known – Standard directory (RFC 5785) for protocol discovery. UCP uses .well-known/ucp.
.well-known is a standard directory (RFC 5785) used for protocol discovery. UCP uses .well-known/ucp to expose merchant capabilities to AI agents.
A
A2A – Agent-to-Agent. Protocol for inter-agent communication so multiple AI agents can coordinate. UCP is compatible with A2A.
Agent2Agent (A2A) refers to protocols and patterns for interactions between AI agents, e.g. when one agent queries another's capabilities or data. UCP is designed to work with A2A where multiple agents participate in a commerce flow.
ACP – Agentic Commerce Protocol. OpenAI's open standard for embedded commerce and Instant Checkout in ChatGPT.
Agentic Commerce Protocol (ACP) is an open standard (Apache 2.0) developed by OpenAI for enabling commerce flows within AI conversations. ACP allows AI assistants to initiate and complete purchases through structured tool calls. See the ACP page for full overview.
Agent Payments Protocol (AP2) – Standard for secure payment credential exchange between agents and merchants, used within UCP.
Agent Payments Protocol (AP2) is a standard for secure payment credential exchange between AI agents and merchants. It is used within UCP to ensure payments initiated by agents are handled safely and in compliance with card network rules.
Agent-Ready – Products and store discoverable and purchasable by AI agents (e.g. via UCP/ACP).
Agent-ready means your products and store are structured and exposed so AI shopping agents can discover, evaluate, and purchase. This typically requires UCP or ACP implementation.
Agentic Checkout – Checkout flows completed by AI assistants on behalf of users.
Agentic checkout is a paradigm where AI assistants complete purchases on behalf of users, reducing friction. The user confirms intent; the AI handles the transaction. Supported by UCP and ACP.
Agentic Commerce – Commerce driven by AI agents acting on behalf of users.
Commerce driven by AI agents acting on behalf of users. The term describes scenarios where AI handles discovery, decision-making, and even purchasing for the user, while the merchant remains the seller of record.
Answer Engine Optimization (AEO) – Optimizing content so answer engines and AI can understand and cite your information. Same as GEO.
Answer Engine Optimization (AEO), historically referred to as Knowledge Engine Optimization, is the practice of optimizing content and site structure so that knowledge engines, AI assistants, and generative search can understand and use your information. Relevant for agentic commerce and citation readiness.
B
Business Agent – Google's term for a brand-specific AI chatbot on Search that can answer questions and handle checkout.
Business Agent is Google's term for a brand-specific AI chatbot on Search that can answer questions about a business and, over time, handle checkout. It aligns with UCP and AI-driven discovery.
C
Capability Discovery – How AI agents discover what commerce capabilities a merchant supports (e.g. via .well-known/ucp).
Capability discovery is the process by which AI agents learn what commerce capabilities a merchant supports. In UCP this is done via .well-known/ucp endpoints.
D
D2C – Direct-to-Consumer. Brands selling directly to customers; need UCP/ACP for agent-ready presence.
Direct-to-Consumer (D2C) brands sell directly to customers without intermediaries. D2C brands need UCP/ACP to become agent-ready for AI shopping.
Delegated Payments Spec – Part of ACP; allows an AI agent to send a payment token to a merchant who then processes payment with their PSP.
The Delegated Payments specification is part of ACP. It allows an AI agent to send a payment token to the merchant, who then processes the payment with their Payment Service Provider (PSP). The merchant remains in control of the transaction.
Direct Offer – Google Ads pilot for special deals surfaced in AI search results.
Direct Offer is a Google Ads pilot that surfaces special deals and offers in AI search results. It is part of the shift toward offers and ads inside AI answers, complementing UCP and Business Agent.
E
ECP – Embedded Checkout Protocol. Extension to UCP for embedding a merchant's checkout UI in an agent for escalations.
Embedded Checkout Protocol (ECP) is an extension to UCP for embedding a merchant's checkout UI within an agent interface when human input is needed (e.g. escalations). It uses JSON-RPC for two-way communication.
F
Feed Hygiene – Quality and accuracy of product data in feeds. Critical for AI agents to parse correctly.
Feed hygiene is the quality, accuracy, and completeness of product data in feeds. Good feed hygiene is critical for AI agents to correctly parse and display products.
G
Generative Engine Optimization (GEO) – Optimization for generative AI outputs. Synonym to AEO.
Generative Engine Optimization (GEO) is a synonym for Answer Engine Optimization (AEO), emphasizing optimization for generative AI outputs and how content is used in AI-generated answers.
I
Instant Checkout – Purchase completion within a single AI conversation, e.g. in ChatGPT.
Instant checkout enables purchases to complete within one AI conversation without leaving the chat interface. Popularized by ACP for ChatGPT.
K
Knowledge Engine Optimization (legacy) – Legacy synonym for Answer Engine Optimization. We now refer to it as AEO.
Knowledge Engine Optimization (KEO) is a legacy synonym for Answer Engine Optimization (AEO). We now standardize on the term AEO but retain this entry for historical context and searchability.
L
llms.txt – A proposed standard text file (like robots.txt) where sites instruct LLMs how to crawl or use content. Relevant for AEO/GEO.
llms.txt is a proposed standard text file (akin to robots.txt) where websites instruct Large Language Models how to crawl or use their content. It is relevant for AEO/GEO readiness and agentic discovery.
M
MCP – Model Context Protocol. Connects AI models to external data and tools; UCP components can have MCP bindings.
Model Context Protocol (MCP) is a protocol for connecting AI models to external data sources and tools. In some contexts (e.g. Google), UCP components like checkout have MCP bindings. It is distinct from commerce protocols like UCP and ACP but can be used alongside them.
Merchant of Record – The entity responsible for the transaction and the customer, even in agentic checkout.
Merchant of Record (MoR) is the entity legally responsible for the transaction and the customer relationship. In UCP/ACP, the original seller remains the Merchant of Record; the AI agent facilitates the purchase on behalf of the user.
P
Product Catalog – Structured product data exposed via UCP or ACP for AI discovery.
A product catalog in agentic commerce is structured product data exposed via UCP or ACP endpoints so AI agents can discover and evaluate products.
PSP – Payment Service Provider. PSPs should implement UCP payment primitives for agentic checkout.
Payment Service Providers (PSPs) process payments for merchants. PSPs should implement UCP payment primitives to enable agentic checkout for their merchants.
S
T
Tokenization (Payment) – Replacing sensitive card data with tokens for secure agent handling. Used in ACP and agentic payments.
Tokenization in payments means replacing sensitive card data with non-sensitive tokens. In ACP and agentic payments, tokens allow an AI agent to authorize a payment without handling raw card details, improving security and compliance.
U
UCP – Universal Commerce Protocol. Open standard for AI-driven discovery, checkout, and post-purchase across surfaces.
Universal Commerce Protocol (UCP) is an open standard that enables AI assistants and shopping agents to discover, evaluate, and purchase products across any merchant or platform. It covers discovery, capability negotiation, cart, payment, and post-purchase. See the UCP page for full overview.
If you'd like us to add a term, contact us.