Design Principle
AI generation needs to show what the system is doing, what users can still do, and how to recover after failure.
Principle
AI generation needs to show what the system is doing, what users can still do, and how to recover after failure.
Design action
Generation areas show stage, queue/time, cancel/retry, partial results, and error recovery; avoid a spinner-only state.
Examples
Positive example: Text, UI, images, reports, code, and workflows are generated asynchronously by AI. Counterexample: A long task only shows Loading.
Apply when
Text, UI, images, reports, code, and workflows are generated asynchronously by AI. Users need to judge state, scope, risk, or next action quickly.
Source notes
Source note: Synthesized from GOV.UK, Carbon, Polaris, Atlassian, Design.Systems, Design Systems Repo, and product-pattern libraries. GOV.UK Design System Patterns: https://design-system.service.gov.uk/patterns/ Carbon Patterns: https://carbondesignsystem.com/patterns/ Shopify Polaris: https://polaris-react.shopify.com/ Atlassian Design System: https://atlassian.design/ Design Systems Repo: https://designsystemsrepo.com/
Agent Directive
Generation areas show stage, queue/time, cancel/retry, partial results, and error recovery; avoid a spinner-only state.