Glossary
The primitives of the agentic web
Definitions for the architectural concepts the portfolio is built on. Each is a primitive that agentic systems will need to name — and a coordinate held in the portfolio.
Agent
A system that perceives, decides, and acts toward a goal with some degree of autonomy.
Attestation
A verifiable statement that something about an agent or its output is true, which another party can check without trusting the source.
Attribution
The source-binding primitive that ties an agentic system's output back to the data, model, and decision that produced it.
Authority
The granted right to act or decide within a scope, the source of an agent's permission.
Autonomous
Acting toward a goal without step-by-step human direction.
Binding
The act of making an agent's commitment enforceable, turning a stated intention into an obligation.
Capability
A scoped, grantable permission to perform a single action, one that can also be revoked.
Constitutional
The foundational, slow-to-change rules an agent operates under, the baseline that every more specific policy must respect.
Convergence
The state in which independent agents arrive at shared agreement, resolving distributed work into one consistent result.
Custody
Controlled holding of an asset or key on another's behalf, responsibility without ownership.
Decision
The discrete choice an agent commits to, the accountable unit of agentic action.
Delegation
The passing of authority from one actor to an agent, so an agent can act for someone else.
Economy
The system of value exchange among agents, defining how autonomous parties price, trade, and settle.
Execution
Where an agent's decision becomes a real-world action with consequences.
Federation
Coordination across independent systems or organizations without surrendering control to a center.
Ghost
An autonomous stand-in that acts on a principal's behalf without their presence.
Governance
The rules and oversight that bound what an autonomous system is allowed to do, with the accountability that comes with it.
Infrastructure
The shared, foundational services autonomous systems depend on to operate.
Integration
The connective layer that lets agentic systems act across the tools and data that already run a business.
Intelligence
The capacity of a system to model, reason about, and act effectively in its domain.
Lineage
The traced path of how data and decisions flow and transform as they move through an agentic system.
Model
The trained artifact that turns data into capability.
Neural
The learned, network-based substrate of modern AI capability.
Observability
The ability to see inside an autonomous system — what it did, why, and whether it is behaving.
Orchestration
The coordination layer that sequences agents, tools, and systems toward an outcome.
Override
Authorized intervention over an agent's action, the sanctioned ability for a trusted party to stop or change it.
Platform
The organizing layer others build on — it defines what a category contains and frames everything beneath it.
Precedent
Prior decisions that guide or bind future ones, the record an agent reasons from to stay consistent.
Process
The executable workflow an agent drives — the ordered sequence of steps, decisions, and handoffs that turns intent into a completed outcome.
Protocol
The rules of engagement by which agents and systems establish identity, authority, and capability.
Provenance
The record of where data, a model, or a decision came from and how it has changed — attribution extended across time.
Quantum
The layer that prepares agentic systems for quantum-era computation and security, covering both the new capabilities and the new threats.
Royalty
A recurring economic claim on the use of an asset or model, earning value from adoption rather than a single sale.
Safety
Bounding what an autonomous system can do so its actions stay within intended limits.
Semantic
The meaning layer of an agentic system, where an agent represents and resolves what things mean rather than how they are spelled.
Settlement
The final, irreversible resolution of an obligation between parties.
Sovereign
Self-governing control over an agent, its data, or its domain, held by the owner rather than an outside platform.
Spatial
The grounding layer that ties agentic action to physical space, giving an agent a model of where things are and how they move.
Substrate
The foundational layer a category is built on — the primitive every higher layer grounds in.
Swarm
The coordination layer for many agents acting as one, where collective behavior emerges from local rules rather than central control.
Temporal
The layer that treats time as first-class, governing how an agent schedules, sequences, and reasons about events.
Treaty
A binding agreement that coordinates multiple autonomous parties under negotiated terms.
Trust
The calibrated confidence that lets one party rely on another's autonomous action, measured rather than assumed.
Twin
A live digital model kept in sync with a real system, giving an agent a faithful counterpart to observe and act on.
Unified
The consolidation of fragmented systems, data, or agents into one coherent surface.