Glossary
Every term used across this site, defined once, so it means the same thing wherever you read it.
Authority
The right, held by an organization, to take a specific action or to delegate that right to someone, or something, else. Organizations already have authority, built for people; PayReality doesn't invent a new authority model for AI, it operationalizes the one that already exists.
Governance
The delegated authority, approval structures, separation of duties, and policies an organization already operates. Governance defines authority; Runtime Authority evaluates it. Neither replaces the other.
Reasoning vs. Authority
An AI agent can reason, analyze, recommend, negotiate, and plan. Authority to actually act has to come from the organization, not from the AI itself, no matter how well it reasoned its way there. Runtime Authority is how that organizational authority gets evaluated at the moment the agent tries to act.
Runtime Authority
The infrastructure layer that determines, immediately before execution, whether an autonomous AI agent's action falls within the authority the organization has already delegated. PayReality's flagship product, built around the authority-continuity thesis: authority may narrow during delegation, but must never expand.
Learn more →AI Workforce Arc
The progression from AI assisting people, to AI executing tasks under supervision, to AI operating as a standing part of the workforce, taking real actions inside enterprise systems on its own. Runtime Authority exists because organizational authority has to hold across all three stages, especially the third.
Authority Graph
The structure modeling who holds delegated authority, who it's been extended to (including AI agents), and what constraints apply, built from an organization's existing governance documents. Authority Graph approvals are versioned snapshots, but a graph-derived policy stays grounded in the specific delegation relationship it was compiled from: that relationship has to still exist and be active at the moment the policy is evaluated, not just at the moment it was reviewed. Revoking or expiring the underlying delegation stops the policy from granting that authority automatically, with no manual policy edit required.
Learn more →AI-Assisted Policy Extraction
AI proposing structured authority (roles, limits, approval chains) from an organization's uploaded governance documents. The AI cannot grant authority or activate a policy on its own: citations and critical values are checked deterministically, human reviewers resolve identities and relationships, high-risk changes require a separate accountable actor to approve, and only an approved, versioned revision can be compiled and activated. Validated so far only in controlled synthetic testing; accuracy against real enterprise documents is a separate evaluation, not yet complete.
Runtime Policies
Deterministic, versioned rules compiled from delegated authority, evaluated against every Intent to determine the specific conditions under which an action is permitted.
Learn more →Intent
A cryptographically signed request an AI agent submits, naming the action it wants to take, before anything executes.
Decision
The result of evaluating an Intent against the Authority Graph and Runtime Policies. Exactly one of three outcomes: Allow, Deny, or Human Review.
Human Review
One of the three Decision outcomes: the Intent is routed to a named, accountable person to resolve, rather than approved or denied automatically.
Evidence Portal
The enterprise evidence layer for Runtime Authority: where every signed Decision is searched, investigated, audited, and exported.
Learn more →Authorization Receipt
A signed, retrievable record of a Decision (actor, request, governing policy version, the human review resolution where one occurred, Capability Authorization state where one exists, and signature), live today via the API and Evidence Portal. A portable, independently verifiable artifact that doesn't require access to PayReality's live systems is still planned, not yet shipped.
Learn more →Execution Receipt
A report of what happened, supplied after the fact by a separately authenticated destination system or a trusted execution adapter, distinct from the Authorization Receipt PayReality issues for its own decision. PayReality verifies the receipt's linkage to the original decision and reconciles it against the action that was actually authorized. A cryptographically authenticated receipt proves what the trusted source reported; it does not independently prove that the underlying business system or real-world event was truthful.
Reconciliation
Comparing a reported Execution Receipt against the action PayReality actually authorized. Resolves to one of six outcomes: MATCHED, MISMATCHED, EXECUTION_FAILED, PARTIAL, RECEIPT_MISSING, or INDETERMINATE.
Capability Authorization
For a decision that needs stronger execution control, a short-lived, single-use authorization PayReality can issue, bound to the exact agent, action, resource, and environment the decision covered. An external, customer-operated enforcement point verifies and consumes it before letting the downstream operation proceed. At most one exists per decision, and a consumed one can't be replayed.
Learn more →Agent
An autonomous AI system registered with its own cryptographic identity, capable of submitting signed Intents on behalf of a Principal.
Principal
The person or role an agent acts on behalf of for policy-evaluation purposes, distinct from who owns or operates the agent day to day.
Delegated Authority
The scope of action an organization has explicitly extended to a person, role, or system: the concept every enterprise already operationalizes for people, and which Runtime Authority extends to AI agents.
Decision Engine
The deterministic evaluation logic that combines the Authority Graph and Runtime Policies into a single Decision. The same Intent, Graph, and Policy always produce the same outcome.
Operator Key
A shared administrative credential that authenticates trusted administrative actions on the platform, such as registering an agent or publishing a policy.
Certificate / Key rotation
The process of replacing an agent's signing key. Rotation cuts off the old key's ability to sign new Intents immediately, without invalidating decisions made under it historically.
Policy compilation
The process of turning authored or extracted policy into deterministic, machine-evaluable rules (using Open Policy Agent / Rego), rather than leaving a written document for a runtime to interpret.