Model Context Protocol (MCP) is an open standard that lets AI agents connect to external tools and data sources. For market research, this means something specific and powerful: AI agents in ChatGPT, Claude, or Cursor can now trigger real consumer studies — recruiting real participants, conducting real conversations, and returning real evidence — not just analyze data that already exists.
This isn’t theoretical. User Intuition’s MCP integration enables a read-write connection between AI agents and a full consumer research platform. The agent doesn’t just search your past research — it creates new research on demand. Studies start at $150, results return in under 3 hours, and every study compounds into a searchable intelligence hub.
What Is MCP (Model Context Protocol)?
MCP is an open standard developed by Anthropic that provides a universal interface for AI agents to interact with external tools. Before MCP, connecting an AI assistant to a specialized platform required a custom API integration for each combination — ChatGPT to Tool A, Claude to Tool A, ChatGPT to Tool B, and so on.
MCP standardizes this. One MCP integration enables any MCP-compatible agent to connect. ChatGPT, Claude, Cursor, and any future MCP-compatible tool can all access User Intuition’s research capabilities through the same standardized interface.
Think of it this way: USB standardized how devices connect to computers. MCP standardizes how AI agents connect to tools. You don’t need a different cable for each device — and you don’t need a different integration for each AI assistant.
User Intuition is the only consumer research platform with a live MCP server. Agents connect via
mcp.userintuition.ai/mcpand can launch preference checks, claim reactions, and message tests with real people from a 4M+ vetted global panel — returning structured results in under 3 hours.
Why MCP Matters for Market Research Specifically?
Most AI tools that touch market research operate in read-only mode. They can:
- Summarize existing transcripts
- Tag themes in uploaded research
- Search across past studies
- Generate reports from existing data
These are valuable capabilities. But they share a fundamental limitation: they only work with data that already exists. If you haven’t studied a topic, the AI has nothing to read.
MCP enables read-write connections. This means AI agents can:
- Design new research studies based on a question
- Recruit real participants from a vetted 4M+ panel
- Conduct AI-moderated interviews with 5-7 levels of laddering
- Return structured findings with evidence trails to real verbatim quotes
The shift from read to read-write is the difference between an AI assistant that helps you process your library and one that can go out and gather new information. (For a deeper look at why this distinction matters, see why AI agents need real consumer data.)
Read vs. Write: Why Most AI Research Tools Only READ
Understanding this distinction is essential for evaluating AI research capabilities:
| Capability | Read-Only Tools | Read-Write (MCP) |
|---|---|---|
| Search past research | Yes | Yes |
| Summarize transcripts | Yes | Yes |
| Tag and theme data | Yes | Yes |
| Create new studies | No | Yes |
| Recruit participants | No | Yes |
| Conduct interviews | No | Yes |
| Generate original findings | No | Yes |
Read-only tools are like a research librarian — they help you find and organize what you already have. Read-write MCP tools are like a research team — they can go create knowledge that didn’t exist before.
When a product manager asks “what do customers think about our new pricing model?”, a read-only tool can search past research for pricing-related findings. A read-write MCP connection can run a new study with real customers and return fresh evidence in hours.
How User Intuition’s MCP Integration Works
The architecture is straightforward:
- AI agent (ChatGPT, Claude, Cursor) receives a research question from the user
- MCP connection translates the request into the research platform’s protocol
- Research platform designs the study, recruits from the 4M+ panel, and conducts AI-moderated interviews
- Findings are structured with evidence trails and returned to the agent
- Agent presents results in the user’s conversation with citations to real quotes
The agent handles the translation between natural language requests (“what do millennials think about sustainable packaging?”) and research parameters (target audience, methodology, sample size). The platform handles everything else — recruitment, moderation, analysis, and evidence structuring.
Security is maintained throughout. Participant data is never exposed to the AI agent. Only structured, anonymized findings — with evidence trails to verbatim quotes — are returned. The platform maintains ISO 27001, GDPR, and HIPAA compliance standards regardless of how the study is initiated.
Three MCP Research Use Cases
1. Preference Checks
Scenario: A product team is deciding between three feature implementations.
Agent prompt: “Run a preference check with 20 enterprise SaaS users — which of these three dashboard layouts do they prefer and why?”
What happens: The agent triggers a study targeting enterprise SaaS users from the panel. Each participant engages in a conversational evaluation of the three options, explaining their reasoning with 5-7 levels of probing depth. Results return with preference rankings and the qualitative reasoning behind each choice.
Time: Under 3 hours. Cost: approximately $400.
Key finding: In a typical 20-person preference check, results include preference rankings with qualitative reasoning, minority dissent with verbatim quotes, and actionable recommendations — all delivered in under 3 hours for approximately $400.
2. Claim Reactions
Scenario: Marketing needs to validate a new value proposition before launching a campaign.
Agent prompt: “Test this claim with 30 B2B procurement managers: ‘Reduce vendor evaluation time by 60% with AI-powered shortlisting.’ How do they react?”
What happens: The agent runs claim reaction interviews where each participant reads the claim and discusses their response — believability, relevance, differentiation from competitors, and what would make it more compelling. Findings include agreement rates, key objections, and suggested improvements, all traced to real quotes.
Time: Under 3 hours. Cost: approximately $600.
3. Message Tests
Scenario: A brand is choosing between email subject lines for a product launch.
Agent prompt: “Test these 4 email subject lines with 40 consumers who have purchased in the last 90 days. Which drives the most interest and why?”
What happens: Each participant evaluates the subject lines in a conversational format, explaining which catches their attention, what it signals to them, and whether they’d open the email. Results include rankings with explanatory themes and minority perspectives.
Time: 3-6 hours. Cost: approximately $800.
ChatGPT + MCP: Running Research from ChatGPT
ChatGPT supports MCP connections through its plugin and integration architecture. With User Intuition’s MCP integration:
Setup: Go to Settings > Connected Apps > Add MCP Server and enter https://mcp.userintuition.ai. OAuth prompts on first use.
Usage: Ask research questions in natural language. ChatGPT translates your request into research parameters, triggers the study, and presents findings directly in the conversation.
Example conversation:
- You: “I need to understand why our premium subscribers are downgrading. Can you run a quick study with 15 recent downgraders?”
- ChatGPT: Triggers a churn study targeting recent downgraders from your CRM, conducts AI-moderated interviews, and returns structured findings with evidence trails.
Best for: Strategists and insights leaders who live in ChatGPT for daily work and want consumer evidence integrated into their AI workflow.
Claude + MCP: Running Research from Claude
Claude has native MCP support, making it the most seamless integration for agentic research:
Setup: Add to Claude Desktop settings or ~/.claude.json: {"mcpServers": {"userintuition": {"url": "https://mcp.userintuition.ai/mcp"}}}. OAuth prompts on first use.
Usage: Claude understands research methodology and can help design study parameters before triggering the study. It can suggest sample sizes, refine target audience criteria, and propose probing strategies — then execute the study and return results.
Example conversation:
- You: “We’re launching a new onboarding flow next month. I want to know if users find it intuitive before we commit engineering resources.”
- Claude: Suggests a 20-person usability study with specific audience criteria, triggers the study, and returns findings with actionable recommendations tied to real user quotes.
Best for: Research teams and strategists who want an AI assistant that understands research methodology and can both design and execute studies. For a step-by-step walkthrough with prompts and cost estimates, see how to run consumer research from Claude Code.
Cursor + MCP: Research Without Leaving Your IDE
This is where MCP gets interesting for product teams. Cursor’s MCP support means developers and product managers can get consumer evidence without switching tools:
Setup: Go to Settings > MCP > Add Server and enter https://mcp.userintuition.ai/mcp. OAuth prompts on first use.
Usage: While working on code or product specs, ask Cursor to validate assumptions with real consumers. Results appear alongside your development work.
Example workflow:
- Developer is building a pricing page
- Asks Cursor: “Quick check — do SMB buyers prefer monthly or annual pricing display? Run a preference check with 15 SMB decision-makers.”
- Cursor triggers the study and returns findings while the developer continues working on other tasks
- Results arrive: “73% preferred annual pricing displayed as monthly equivalent, primarily because it made the cost feel comparable to their existing tools”
Best for: Product teams who want to embed consumer evidence into sprint cycles without the overhead of scheduling research through a separate team. See also: agentic research for product teams.
Security and Compliance
MCP-powered research maintains the same security and compliance standards as direct platform access:
- GDPR-compliant — participant data protection and consent management
- ISO 27001-aligned — information security management
- HIPAA-aligned — healthcare-related research meets regulatory requirements
- SOC 2 Type II examination underway — additional security assurance
Data handling principles:
- Participant personal data is never exposed to the AI agent
- Only structured, anonymized findings are returned through MCP
- Evidence trails reference anonymized participant IDs, not personal information
- Study data is stored in User Intuition’s secure infrastructure, not in the AI agent’s context
- All participant consent is managed by the platform, not the agent
MCP Configuration: What Setup Looks Like
Use the hosted endpoint, https://mcp.userintuition.ai/mcp, with OAuth in compatible clients. For local stdio, run npx -y @userintuition-ai/mcp with USERINTUITION_API_KEY set to your ui_sk_ key. The CLI supports browser login and API-key management. Hosted OAuth does not require exchanging a user-supplied API key.
For a local stdio client that uses an MCP configuration file:
{
"mcpServers": {
"userintuition": {
"command": "npx",
"args": ["-y", "@userintuition-ai/mcp"],
"env": {"USERINTUITION_API_KEY": "ui_sk_your_key_here"}
}
}
}Keep real keys out of version control. For shell workflows:
npm install -g @userintuition-ai/mcp
userintuition-mcp login
userintuition-mcp list
userintuition-mcp list_studiesStart with a read such as list_studies to check your connection. See the current MCP setup guide for client-specific instructions and the research skills for workflow guidance.
What Is the Research Lifecycle Through MCP?
Create a metadata draft with create_study, then send the research brief through customize_study. Relay any planning questions to the user. Retrieve the persisted plan with get_study and obtain approval before recruitment. The user must choose panel or BYOP explicitly.
For a panel study, use launch_panel with dry_run: true to obtain the recruitment estimate. Show the country, language, cost, and timeline; launch with the same settings after approval. Each launch specifies one country. Audiences below 10% incidence require a feasibility request.
For a BYOP study, use create_participants with 1–100 unique participant emails per batch after the saved plan and invitations are approved. Invitations send by default; set silent: true on individual participant records when invitations should not send. Source customer lists through your own authorized export or integration; MCP has no direct CRM segment-sync tools.
get_study_report returns report text, references, interview count, stale status, and timestamps in JSON. Use generate_report when analysis is needed, and get_interview to verify supporting messages and recording links. Preference shares, credibility scores, and ranked themes are not guaranteed typed fields in this response.
Agents can list studies, retrieve selected study reports, and inspect supporting interviews. They can compare that retrieved evidence in their own context. Direct cross-study Intelligence Hub search is available in the dashboard; the current public API and MCP catalog do not expose Hub search, Hub sessions, or presentation generation.
Getting Started: Connecting Your First AI Agent
Step 1: Choose your AI tool. ChatGPT, Claude, and Cursor all support MCP. Choose whichever you use most in your daily workflow.
Step 2: Configure the MCP connection. Add User Intuition as an MCP source in your tool’s settings. Configuration takes under 5 minutes using the examples above.
Step 3: Run your first study. Start with a simple preference check — 10-15 participants, one clear question. See how fast real evidence arrives in your AI conversation.
Step 4: Scale up. As you see the speed and quality of results, expand to claim reactions, message tests, and larger panels. Build the habit of validating assumptions with real consumer evidence before committing resources.
Step 5: Compound. Save relevant study IDs and compare reports as new evidence arrives. Use the dashboard Intelligence Hub for native cross-study search.
The organizations connecting AI agents to real consumer research via MCP are building a structural advantage: decisions backed by real evidence, validated in hours, compounding over time. The tools are available now. The question is whether you start building that advantage today or let competitors build it first.
Related: Consumer Research API: How AI Agents Call Studies Programmatically | How to Run Consumer Research from Claude Code | Agentic Consumer Insights Research Guide
Ready to connect your AI agent to real consumer research? Explore the agentic research platform, read the MCP server docs, or sign up free (3 interviews, no credit card).