Overview
How might we design proactive nudge so people can trust and act on AI output?
When this pattern fits
- Perfect for writing tools, productivity applications, and platforms where proactive, non-intrusive suggestions help users discover features and improve their work.
When to skip or lighten it
- Users in focused flow who dismissed help repeatedly.
- High-stakes screens where any popup causes errors.
- Suggestions with low confidence that add noise.
States
State model coming soon
Key UX elements
Key UX elements coming soon
Anti-patterns to avoid
Modal nudges on every session start.
Nudge with no dismiss or snooze forever.
Same tip after user already used the feature.
Nudges that block primary action buttons.
How products use it
| Product | Implementation |
|---|---|
| Grammarly | Inline suggestions while typing without opening a panel. |
| Microsoft Word | Editor nudges for clarity and style. |
| Google Docs | Smart compose and assist chips in context. |
| Notion | AI assist prompts when selecting blocks or empty pages. |
Implementation
Copy this prompt to generate a production-ready implementation in Cursor, Claude Code, Lovable, or any AI coding agent.
Generate a production-ready implementation of the "Proactive Nudge" AI interface design pattern.
Pattern Definition:Frequently asked questions
Nudge vs contextual AI suggestions?
Nudges are lighter hints. Contextual suggestions propose a specific AI action with a button.
How often?
Cap frequency per feature and respect dismissals across sessions.
Measure success?
Accept rate and task completion lift, not just impressions.
Accessibility?
Nudges must not steal focus; offer keyboard dismiss.