Overview
How might we design progressive disclosure so people can trust and act on AI output?
When this pattern fits
- Perfect for long-form content, research reports, and complex AI outputs where users need both high-level summaries and detailed information on demand.
When to skip or lighten it
- Safety-critical details that must be visible before consent or execute (do not hide risk).
- Expert-only consoles where collapsing essentials adds clicks without helping novices.
- Tiny answers where “show more” is pure chrome with no extra content.
States
State model coming soon
Key UX elements
Key UX elements coming soon
Anti-patterns to avoid
Hiding irreversible side effects behind Show more or Advanced.
Endless nested accordions that bury the primary answer.
Different disclosure rules on mobile vs desktop with no shared hierarchy.
Default-collapsed citations or errors that users need to trust the claim.
How products use it
| Product | Implementation |
|---|---|
| ChatGPT | Compact replies with expandable reasoning, sources, or tool details. |
| Claude | Artifacts and extended thinking open from the thread when needed. |
| Perplexity | Answer first; research steps and Links available as deeper layers. |
| Gemini | Overview-style answers with expandable related content and tools. |
Real-world examples
How shipped products implement progressive disclosure, from our teardown guides.
ChatGPTChatGPT personalization UX: tone presets, memory & profile design
Personalization & memory
ClaudeClaude personalization UX: profile, memory & voice design
Personalization & memory
Google MapsAsk Maps by Google Maps: chat beside the map
Local search & planning
GrokGrok composer UX: skills, models & voice input
Design the composer
KimiKimi composer UX: agent pills, Instant/K3 & skills
Design the composer
ManusManus composer UX: assign tasks, modes & agent tiers
Design the composer
ManusManus onboarding UX: role segmentation & Pro upsell
Getting to first value
ManusManus output UX: agent runs, files & deliverables
Output, artifacts & refinement
PerplexityPerplexity personalization UX: vertical profiles, response style & memory
Personalization & memory
QwenQwen composer: a quiet bar for a multimodal studio
Design the composer
QwenQwen output: keep the chain in the thread
Output, artifacts & refinement
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 "Progressive Disclosure" AI interface design pattern.
Pattern Definition:Frequently asked questions
What is progressive disclosure in AI UX?
Progressive disclosure shows a simple default (usually the answer), then lets users expand sources, steps, tools, or advanced controls when they need more depth.
Should citations be progressive or always visible?
Keep at least lightweight provenance visible on claims users must verify. Deeper source lists can progressive-disclose; hiding all evidence by default weakens trust.
How does it differ from progressive feature unlock?
Progressive disclosure hides UI complexity on a single screen. Progressive feature unlock gates capabilities over time as the user learns or upgrades.
When is progressive disclosure unsafe?
When permissions, spend, deletions, or legal notices are tucked behind Advanced. Anything authorizing a side effect belongs above the fold.