Skip to main content
The User Intuition MCP server covers two distinct research domains, each with its own set of study modes.

​
Human Signal modes (paid panel)

Use ask_humans to run a study against the RepData paid panel. Each respondent costs roughly 1–1–5. Pass the mode parameter to select the study type.
ModeWhen to useConstraints
preferenceCompare 2–5 options and find out which resonates mostRequires 2–5 options
claimTest whether a belief or positioning statement is convincingSingle claim
messageTest message clarity — does your copy land?Single message
Results are retrieved with get_results. Studies can be modified while in creating, queued, or fielding status via edit_study, and cancelled at any point with cancel_study.

​
Research-platform modes (your own panel / customers)

For live AI-moderated interview studies, mode is expressed as flags on the assistant record rather than a separate field. See the Study-creation playbook for full flag semantics.
Modeenable_chatenable_video_recordingNotes
Voice (default)falsefalseAudio interview, no video
ChattruefalseText-chat interview
VideofalsetrueVoice interview with video recording
Chat + videotruetrueChat interview with video recording
Both flags default to false. You can combine them freely.

​
Concept tests

Concept tests are a special study variant available in both voice and chat modes.
  1. Set is_concept_test: true on the assistant when calling create_assistant or update_assistant.
  2. Use upload_concept_link to attach concept material (a URL, mockup, or prototype link). The tool deduplicates by label, so re-uploading the same label updates the URL in place rather than creating a duplicate entry.
  3. Remove a concept link with delete_concept_link by passing its id.
Before creating a concept-test study, load the matching prompt from the server:
Study variantPrompt to use
Voice concept testconcept_test_planning_context
Chat concept testconcept_test_chat_planning_context
See the Prompts page for the full prompt reference.