Back to compare picker

Prompt suggestions vs Search suggestions vs Command palette

Prefer prompt suggestions when the next step is an editable natural-language AI request and the user needs examples of supported capabilities.

Decision dimensions

Dimension Prompt suggestionsSearch suggestionsCommand palette
UI or UX UI + UX - AI prompt starter surfaceUI + UX - Suggested-query surfaceUI + UX - Modal command surface
UI guidance Render contextual prompt starter cards or chips with a task label, editable prompt text, capability boundary, selected state, disabled or stale state, and explicit send action.Render a search field with query suggestions, active option, selected suggestion, and no-suggestion state.Render a compact dialog-like command surface with a search input, current scope, typed command mode, active result, command metadata, and empty state.
UX guidance Help users discover useful AI requests while preserving control over the exact prompt that is sent.Help users formulate better search queries without forcing a choice.Accelerate expert navigation and repeated actions across a large product while preserving ordinary navigation for novice and low-frequency users.
Good UI Each prompt starter shows an action label, full prompt preview, and boundary note such as uses selected thread only.Suggestions appear directly below the search field with active row, readable labels, and optional matched text.Centered command surface with input, shortcut hint, scope chip, grouped commands, command type labels, and a visible active row.
Bad UI Three tiny chips labeled summarize, improve, and more with no context or affordance.Suggestions overlay unrelated content without a dismiss path.Huge branded modal that buries the input below decorative content.
Good UX Selecting a starter loads a draft, moves focus to the prompt editor, and lets the user adjust scope before sending.Users can keep typing, choose a suggestion, dismiss suggestions, or submit their own query.Keyboard shortcut or visible trigger opens the palette, focus lands in the command input, arrows move the active row, and Enter activates the highlighted safe command.
Bad UX Clicking a suggestion immediately runs a consequential task without review.Auto-submitting the top suggestion.Palette is the only way to reach important navigation.
Best fit The AI surface supports open-ended requests and users need examples of useful, supported tasks.The system can predict likely queries.Users need to traverse a broad product surface quickly.
Avoid when The user must provide fixed required fields that are clearer as a form.Suggestions are low quality or biased toward irrelevant content.The app has only a few obvious actions.
Required state Initial state with contextual prompt starters and capability or source boundary notes.No input state.Closed state with discoverable trigger.
Accessibility burden Expose suggestions as named buttons, list items, or cards with clear keyboard focus and selected state.Expose the suggestions list and active option.Use dialog semantics with a clear name and modal behavior when the rest of the page is inert.
Common misuse Showing generic chips such as summarize, improve, or make better with no object-specific context.Auto-submitting the top suggestion without confirmation.Hiding basic navigation behind a keyboard-only palette.

Prompt suggestions

UI or UX
UI + UX - AI prompt starter surface
UI guidance
Render contextual prompt starter cards or chips with a task label, editable prompt text, capability boundary, selected state, disabled or stale state, and explicit send action.
UX guidance
Help users discover useful AI requests while preserving control over the exact prompt that is sent.
Good UI
Each prompt starter shows an action label, full prompt preview, and boundary note such as uses selected thread only.
Bad UI
Three tiny chips labeled summarize, improve, and more with no context or affordance.
Good UX
Selecting a starter loads a draft, moves focus to the prompt editor, and lets the user adjust scope before sending.
Bad UX
Clicking a suggestion immediately runs a consequential task without review.
Best fit
The AI surface supports open-ended requests and users need examples of useful, supported tasks.
Avoid when
The user must provide fixed required fields that are clearer as a form.
Required state
Initial state with contextual prompt starters and capability or source boundary notes.
Accessibility burden
Expose suggestions as named buttons, list items, or cards with clear keyboard focus and selected state.
Common misuse
Showing generic chips such as summarize, improve, or make better with no object-specific context.

Search suggestions

UI or UX
UI + UX - Suggested-query surface
UI guidance
Render a search field with query suggestions, active option, selected suggestion, and no-suggestion state.
UX guidance
Help users formulate better search queries without forcing a choice.
Good UI
Suggestions appear directly below the search field with active row, readable labels, and optional matched text.
Bad UI
Suggestions overlay unrelated content without a dismiss path.
Good UX
Users can keep typing, choose a suggestion, dismiss suggestions, or submit their own query.
Bad UX
Auto-submitting the top suggestion.
Best fit
The system can predict likely queries.
Avoid when
Suggestions are low quality or biased toward irrelevant content.
Required state
No input state.
Accessibility burden
Expose the suggestions list and active option.
Common misuse
Auto-submitting the top suggestion without confirmation.

Command palette

UI or UX
UI + UX - Modal command surface
UI guidance
Render a compact dialog-like command surface with a search input, current scope, typed command mode, active result, command metadata, and empty state.
UX guidance
Accelerate expert navigation and repeated actions across a large product while preserving ordinary navigation for novice and low-frequency users.
Good UI
Centered command surface with input, shortcut hint, scope chip, grouped commands, command type labels, and a visible active row.
Bad UI
Huge branded modal that buries the input below decorative content.
Good UX
Keyboard shortcut or visible trigger opens the palette, focus lands in the command input, arrows move the active row, and Enter activates the highlighted safe command.
Bad UX
Palette is the only way to reach important navigation.
Best fit
Users need to traverse a broad product surface quickly.
Avoid when
The app has only a few obvious actions.
Required state
Closed state with discoverable trigger.
Accessibility burden
Use dialog semantics with a clear name and modal behavior when the rest of the page is inert.
Common misuse
Hiding basic navigation behind a keyboard-only palette.
Decision rules
  • Prefer prompt suggestions when the next step is an editable natural-language AI request and the user needs examples of supported capabilities.
  • Prefer search suggestions when the next step is a query for content retrieval, results, or browsing and the user can still submit a free-form query.
  • Prefer a command palette when selecting a row executes an app command, opens a destination, changes mode, or performs an action with a defined scope.
  • Do not use prompt suggestions for fixed required inputs; use a form or structured workflow when the task needs specific fields, validation, or approval.
  • Do not use a command palette for AI prompt starters unless each row's consequence, scope, and side effect are explicit before activation.
  • If selecting text would immediately run an AI task, label it as an action and add review or approval for consequential output; if it loads editable text, label it as a prompt starter.
  • Prompt suggestions must disclose capability boundaries such as selected text required, available sources, unsupported languages, or missing permissions.
  • Search suggestions can be stale or low confidence without being dangerous; prompt suggestions that overpromise model access, grounding, or certainty can produce misplaced trust and should be hidden or rewritten.
Inspect live examples
Failure modes
  • Using generic prompt chips that could belong to any AI product and reveal no supported capability.
  • Auto-submitting a suggested prompt before the user can edit missing context.
  • Treating an app command such as delete workspace as a harmless suggested prompt.
  • Using search suggestions for known-value completion where autocomplete would be clearer.
  • Showing saved or shared prompts after the model, agent tools, source access, or permissions changed.