Agent

What it means

The model decides sequences of actions (search, draft, click, call APIs) within policies until the task completes or hits a human checkpoint.

Why designers should care

Users must see what the agent did, what it will do next, and how to stop it. Opaque autonomy erodes trust fast after one wrong action.

Example

A “Prepare release notes” agent gathers merged PRs, drafts bullets, opens a doc, and pauses for PM approval before posting to Slack, with each step logged in a timeline.

Common mistakes

• • Full autonomy on irreversible actions (send, delete, purchase) without confirmation.
• • No activity log when agents touch multiple systems.
• • Marketing “agent” labels on simple single-shot chat features.

Related terms

• Tool Use: The model emits structured calls (name + arguments) that your app executes, then feeds results back into the conversation for the next step.
• Human-in-the-Loop: The product pauses at defined checkpoints for human judgment (approve send, merge PR, publish article, charge card), even if AI drafted the action.
• Workflow: Defined stages (inputs → generate → review → export) often spanning people, systems, and multiple model calls with clear deliverables.
• Prompt Injection: Malicious or accidental instructions embedded in retrieved or pasted content override intended rules (“ignore previous instructions and…”).
• Latency: Network, queue, retrieval, tool calls, and generation length all add wait time before users can read, edit, or act on output.

Explore more

• Agentic UX framework