The language of agentic work.

Work that a responsible person or system has judged complete against the result that was promised.

The software environment around a model that supplies it with context, memory, tools, permissions and an execution loop, and that constrains what its choices are able to affect.

A metadata surface where agent or service records can be published and found, typically carrying a name, endpoint, authentication requirement, declared skills and schemas.

The process and infrastructure that executes an agent's work — starting it, allocating workers, enforcing resource limits, streaming output and terminating it.

Describes a system or service designed with software agents as primary users or participants.

Describes a system that supports a specified interaction by a software agent.

An open protocol that lets independently operated agent systems discover one another, exchange messages and track work as a task with its own state and output artefacts.

Describes AI software that chooses and carries out steps towards a goal rather than following only a predetermined sequence.

Buying and selling in which software agents participate in commercial choices or actions for people or organisations.

Economic activity in which AI agents help people and organisations produce goods, deliver services, coordinate work, and buy and sell.

The broad field of systems that infer from information how to produce predictions, content, recommendations or decisions towards objectives.

An AI system that can select steps towards a goal and carry out actions through available tools.

An AI application through which a person asks for help with information or work.

The learned component of an AI system that maps supplied information to predictions or other outputs.

Describes a product, service or business designed around AI as a central part of how it delivers value.

Use of AI to support a person’s work while the person remains actively involved.

Permission to take specified actions on someone’s behalf, within stated limits.

Work carried out by a system with reduced human action, using fixed rules, AI or a combination of both.

The degree to which a system can learn or act without human involvement after people delegate work to it.

A shared message and state contract for selected buyer, seller or platform interactions, typically covering how a checkout is formed, updated and turned into a merchant order.

The selected information available to an agent during a current decision — the objective, the policy, the conversation so far, the business records, the retrieved material and the working notes.

A bounded job a person or organisation authorises software to carry forward: an objective, the context and authority to pursue it, the action taken, the evidence left behind, and somebody who stays accountable for the result.

Assignment of work to a person or system, together with the decisions they are permitted to make.

Measurement of a system's behaviour against a defined test — covering the model's output, the sequence of steps it chose, and the effect it had outside itself.

AI that creates synthetic content, such as text, images or audio, using patterns learned from data.

The property that repeating the same request produces the same single effect, rather than a second one — usually achieved by attaching a stable key to the logical action.

A record of bounded intent or authority, represented so that another system can verify it — naming who authorised what, for whom, within which limits and for how long.

Retained state an agent can use in a later session or job — a stored preference, a prior job's outcome, or a summary extracted from an earlier interaction.

The legal entity that sells to the customer in a transaction — the party responsible for the sale, its tax treatment, refunds, chargebacks and customer support.

An open protocol that gives an AI application a standard way to discover and call external tools and data sources, using a host, client and server arrangement.

Recording enough state and activity to inspect a run afterwards — the model calls, retrievals, tool executions, approvals, costs, errors and external effects that made up a job.

The state transitions, scheduling, branching, waiting, retry, cancellation and compensation that coordinate multi-step work.

The network or account system that actually moves value — a card network, a bank transfer system, a wallet or a blockchain.

The person or organisation on whose behalf software acts — whose authority, funds and obligations sit behind the action.

The record of where information came from, how it changed, and which activity or actor produced it.

The process that finds external records and supplies them to a model as context — commonly shortened to RAG, for retrieval-augmented generation.

Work performed for another party, such as transport, teaching, equipment maintenance or bookkeeping.

The point at which a payment method says the seller has received final or usable value. What counts as final differs by method.

A bounded piece of work contributing to a larger objective.

A function an AI system can use to obtain information or take an action.

A connected arrangement of tasks and decisions through which work proceeds.