Agentic AI Glossary: Key Terms & Concepts

An audit trail is a step-by-step sequential record that offers evidence of a financial transaction's documented history back to its source. With the use of the audit trail, an auditor can ...read more

The approval workflow is a standard corporate procedure in which data must be signed off at a given point. Approval procedures necessitate various checks from a range of stakeholders throu...read more

API integration is a feature that allows different software systems, platforms, or applications to seamlessly communicate with each other. It enables the exchange of data, functionality, a...read more

Task automation is a highly useful feature that allows users to streamline and automate their repetitive tasks and processes, saving them significant amounts of time and effort. This featu...read more

Parallel execution is a software feature that allows multiple tasks or processes to be executed simultaneously. This means that instead of completing one task and then moving on to the nex...read more

Workflow automation is a key feature in software that streamlines and automates repetitive tasks and processes, improving efficiency, accuracy, and productivity across an organization. It ...read more

Context management in AI code generators refers to the system's ability to understand, retain, and apply relevant information throughout the code generation process. This feature enables t...read more

Coordinating several AI agents so they work on parts of a larger task and hand results to one another, instead of relying on a single model call. The orchestration layer decides which agen...read more

A language model is given descriptions of the functions or tools available to it and, when a request needs one, returns a structured call naming the function and its arguments instead of p...read more

Programmatic control of a real web browser, so a tool can navigate pages, fill forms, click elements, and read what is rendered. In AI coding tools it is what lets an agent open the app it...read more

An agent runtime is the process that actually executes an agent: it holds the loop that calls a model, reads the response, runs whatever action the model asked for, feeds the result back, ...read more

Agent memory is what an agent still knows at the start of the next run: facts it was told, decisions it made, and a record of what it has already done for this user. Without it every conv...read more

Durable execution means the state of a run is written down at each step, so a run that dies partway through can carry on from the last completed step rather than starting again. It matter...read more

Sandboxed execution gives an agent an isolated machine, container or browser of its own to work in, so code it writes and pages it opens cannot reach anything else. This is what makes it ...read more

Agent evaluation scores what an agent produced against a fixed set of test cases, so a prompt change, a model swap or a new action can be compared with the version already in production. ...read more