Skip to main content
COMPEL Body of Knowledge
Get Certified Platform

COMPEL Glossary / computer-use-browser-use-agent

Computer use / browser-use agent

An AI agent that operates through a browser or computer user interface against third-party systems rather than via APIs — taking screenshots, clicking, typing, and scrolling.

What this means in practice

Anthropic Computer Use (October 2024) and OpenAI Operator (January 2025) are canonical examples; governance concerns include credential handling, blast-radius, cross-site consent, and provider-side audit trail.

Synonyms

browser-use agent , computer-use agent , GUI-operating agent

Related Terms Network

This diagram shows how Computer use / browser-use agent connects to other concepts in the COMPEL body of knowledge. Each linked node is a term you can explore in turn.

Related terms for Computer use / browser-use agent Network graph showing 3 related glossary terms connected to Computer use / browser-use agent. Computer use /… Sandboxing Authority … Tool use /…
3 related concepts connected to Computer use / browser-use agent.

The COMPEL Glossary Graph visualizes relationships between framework terminology, showing how concepts interconnect across domains, stages, and pillars. Term nodes cluster by pillar affiliation while cross-references reveal semantic dependencies — for example, how risk appetite connects to control effectiveness, model governance, and assurance requirements. This network representation helps practitioners navigate the framework vocabulary and understand that COMPEL terminology forms a coherent conceptual system rather than isolated definitions.

Accessible list of related terms
  • Sandboxing — Technical isolation that limits an agent's actions to a bounded environment — file-system jail, network-namespace restriction, container, VM, or ephemeral cloud workspace — so that a misbehaving agent cannot reach systems of record, production data, or external networks.
  • Authority chain — The traceable lineage from an organization's decision-rights authority — through any delegating humans — to the agent executing the action.
  • Tool use / function calling — A prompt pattern authorizing the model to request named functions with structured arguments — searching the web, reading a database, calling a calculator, triggering an API — rather than generating all answers from its weights.

See also

  • Sandboxing — Technical isolation that limits an agent's actions to a bounded environment — file-system jail, network-namespace restriction, container, VM, or ephemeral cloud workspace — so that a misbehaving agent cannot reach systems of record, production data, or external networks.
  • Authority chain — The traceable lineage from an organization's decision-rights authority — through any delegating humans — to the agent executing the action.
  • Tool use / function calling — A prompt pattern authorizing the model to request named functions with structured arguments — searching the web, reading a database, calling a calculator, triggering an API — rather than generating all answers from its weights.