Progressive Disclosure

What it means

The UI defaults to the smallest useful answer surface, with explicit affordances (expand, show steps, view sources) for deeper content.

Why designers should care

Long LLM replies overwhelm users. Disclosure patterns pair well with chain-of-thought, citations, and tool logs without dumping everything at once.

Example

A research answer shows three bullet takeaways; “Show evidence” expands cited quotes, and “Show agent steps” opens the tool timeline for power users.

Common mistakes

• • Hiding critical warnings or citations behind too many clicks.
• • Expand-all by default, defeating the purpose of layering.
• • No persistent expanded state when users navigate away and return.

Related terms

• Chain of Thought: The model is instructed to think step-by-step (often internally or visibly), improving accuracy on multi-step tasks at the cost of length and latency.
• Explainability: Explainability covers citations, chain-of-thought summaries, highlight-to-source links, confidence labels, and plain-language “why this recommendation” copy.
• Streaming Response: The client renders partial text as it arrives over HTTP or WebSocket until the model signals completion or stop.
• Prompt: Prompts set role, task, constraints, format, and context; the model’s reply is a direct response to that combined instruction stack.
• Agent: The model decides sequences of actions (search, draft, click, call APIs) within policies until the task completes or hits a human checkpoint.