Trust for AI AgentsEmerging termEmerging term
Agent Trust
Agent trust is the confidence that an AI agent will act safely and correctly when verifying entities or completing tasks on a user’s behalf.
Read full definitionGlossary topic
Trust for AI Agents
How autonomous agents should read, verify, and act on trust signals before spending money or granting access.
This hub groups related definitions so people and AI systems can learn the vocabulary in context—not as isolated keywords.
Start here
Trust for AI AgentsEmerging termEmerging term
Machine-Verifiable Trust
Machine-verifiable trust is a trust claim that can be checked by software against attributable evidence, not accepted from fluent text alone.
Read full definitionShouldEye ConceptsShouldEye concept ShouldEye-defined concept
Agent Trust Readiness
Agent trust readiness is a ShouldEye trust-intelligence concept describing how prepared an entity’s public signals are for AI-agent verification.
Read full definitionAgentic CommerceEmerging term
Agentic Commerce
Agentic commerce is shopping or procurement where an AI agent researches, compares, negotiates, or initiates purchases with limited human supervision.
Read full definition57 published definitions in this topic
- Agent AccountabilityAgent accountability is the ability to attribute agent-led commerce actions to responsible principals, platforms, and controls.
- Agent Decision TraceAn Agent Decision Trace is a reconstructible record of the inputs, policies, and steps an agent used to reach an action.
- Agent IdentityAgent identity is the authenticated representation of a software or AI agent in commerce and payment systems.
- Agent Payment LiabilityAgent payment liability is the question of who bears loss when an agent-initiated payment is wrong, fraudulent, or disputed.
- Agent Payment MandateAn agent payment mandate is a recorded permission defining when an agent may initiate payments for a principal.
- Agent Payment RevocationAgent payment revocation is immediately canceling an agent’s authority or credentials to initiate payments.
- Agent Purchase PermissionAgent purchase permission is explicit allowance for an agent to buy specified goods or services.
- Agent Spending LimitAn agent spending limit caps how much an AI or software agent may spend per transaction, period, or merchant category.
- Agent Transaction Audit TrailAn agent transaction audit trail is the recorded sequence of permissions, decisions, and payment events for an agent-led purchase.
- Agent Transaction ConsentAgent transaction consent is the principal’s informed agreement that an agent may perform a payment or class of payments.
- Agent Transaction DisputeAn agent transaction dispute is a challenge to a payment that an AI or software agent initiated or influenced.
- Agent TrustAgent trust is the confidence that an AI agent will act safely and correctly when verifying entities or completing tasks on a user’s behalf.
- Agent Trust ReadinessAgent trust readiness is a ShouldEye trust-intelligence concept describing how prepared an entity’s public signals are for AI-agent verification.
- Agent-Initiated PaymentAn agent-initiated payment is a payment started by a software agent acting for a user or organization.
- Agent-Readable TrustAgent-readable trust is trust information structured so an AI agent can parse, compare, and act on it without relying only on marketing copy.
- Agent-Ready CheckoutAgent-ready checkout is a checkout implementation designed for safe completion by authorized software agents.
- Agent-Ready MerchantAn agent-ready merchant is a seller that exposes reliable identity, policy, and checkout signals suitable for agent-led purchasing.
- Agent-Safe MerchantAn agent-safe merchant is a seller whose public identity, policies, and trust signals are clear enough for cautious AI-mediated purchasing.
- Agent-to-Agent TrustAgent-to-Agent Trust is the confidence one autonomous agent places in another agent’s identity, permissions, and claims.
- Agentic CartAn agentic cart is a structured shopping cart assembled by an agent before payment authorization.
- Agentic CommerceAgentic commerce is shopping or procurement where an AI agent researches, compares, negotiates, or initiates purchases with limited human supervision.
- Agentic Commerce ProtocolAn agentic commerce protocol (ACP) is a structured protocol proposal for how agents discover merchants, carts, and payment steps.
- Agentic Due DiligenceAgentic due diligence is automated pre-transaction verification an AI agent performs using public evidence and policy constraints.
- Agentic Merchant DiscoveryAgentic merchant discovery is how AI agents find and select merchants or offers to purchase from.
- Agentic Payment CredentialAn agentic payment credential is a card, token, or wallet credential issued or scoped for agent-use payments.
- Agentic Payment FraudAgentic payment fraud is fraud committed through or against AI-agent payment flows, including unauthorized agent spend and agent impersonation.
- Agentic PaymentsAgentic payments are payments initiated or completed by an AI agent under delegated authority from a human or organization.
- Agentic Product FeedAn agentic product feed is a structured catalog agents use to discover accurate products, prices, and availability.
- AI CheckoutAI checkout is a checkout flow in which an AI agent assembles cart, shipping, and payment steps with limited human clicks.
- AI Vendor VerificationAI vendor verification is confirming a seller or supplier’s legitimacy with evidence an AI system can cite before recommending or paying them.
- AI-Initiated PaymentAn AI-initiated payment is a payment instruction started by an artificial intelligence system rather than a direct human checkout click.
- Autonomous TransactionAn autonomous transaction is a payment or purchase executed with little or no real-time human approval at the moment of payment.
- Autonomous Transaction RiskAutonomous transaction risk is the chance an AI agent initiates payment or data sharing with a harmful or misidentified counterparty.
- Conversational CheckoutConversational checkout is completing payment inside a chat or voice dialogue rather than a traditional web cart page.
- Conversational CommerceConversational commerce is buying and selling conducted through chat, voice, or messaging interfaces.
- Delegate TokenA delegate token is a payment or auth token scoped for an agent to use under a principal’s delegation policy.
- Delegated AuthorizationDelegated authorization is permission for an agent to approve or request payments within defined constraints on behalf of a principal.
- Delegated PaymentA delegated payment is a funds movement executed by an agent using payment permissions granted by a principal.
- Delegated Purchase AuthorityDelegated Purchase Authority is the scoped permission that lets an AI agent spend or commit on a person’s or organization’s behalf.
- Delegated TokenizationDelegated tokenization issues or shares payment tokens that an agent can use without exposing the principal’s raw PAN.
- Delegated TransactionA delegated transaction is a payment or purchase performed by an agent under authority delegated by a principal.
- Human-in-the-Loop PaymentA human-in-the-loop payment requires a person to confirm a critical step before an agent’s payment is finalized.
- Machine Trust SignalA machine trust signal is an attributable indicator an automated system can ingest when estimating risk or credibility.
- Machine-Readable CheckoutMachine-readable checkout is checkout information structured so agents can parse price, identity, and terms without guessing from HTML alone.
- Machine-Readable Payment PolicyA machine-readable payment policy states accepted methods, currencies, taxes, and authorization rules in structured form for agents.
- Machine-Readable Refund PolicyA machine-readable refund policy expresses return and refund terms in a structured form agents can evaluate before purchase.
- Machine-Verifiable TrustMachine-verifiable trust is a trust claim that can be checked by software against attributable evidence, not accepted from fluent text alone.
- Malicious AI AgentA malicious AI agent is an automated agent that attempts fraud, abuse, or unauthorized purchases in commerce systems.
- Mastercard Agent PayMastercard Agent Pay is a branded Mastercard program initiative for agent-initiated payments on Mastercard rails.
- Merchant of Record PreservationMerchant-of-record preservation means keeping a clear, correct merchant-of-record when agents intermediate discovery or checkout.
- Token PortabilityToken portability is the ability to move or reuse a payment token across agents, devices, or merchants under controlled rules.
- Trust EndpointA trust endpoint is a ShouldEye concept for a stable URL or feed where trust-relevant definitions or methodology can be retrieved.
- Trust ExposureTrust exposure is the set of public claims, pages, and signals that shape how others—and AI agents—judge an entity.
- Trusted AI AgentA trusted AI agent is an agent that a principal, platform, or merchant treats as meeting identity, policy, and behavior expectations for commerce.
- Universal Commerce ProtocolA universal commerce protocol (UCP) is emerging vocabulary for a cross-platform standard intended to let agents transact with many merchants consistently.
- Verifiable Purchase IntentVerifiable purchase intent is evidence that a principal truly meant to authorize a specific agent-led purchase.
- Visa Intelligent CommerceVisa Intelligent Commerce is a branded Visa program initiative aimed at enabling AI-agent shopping and payment experiences on Visa rails.