Your research may finish in Wynter, but the work that follows often happens in an AI assistant. A product marketer starts a brief in ChatGPT. An analyst explores patterns in Claude. A technical marketer works in Codex or Claude Code.
Without a connection, the evidence has to be copied, exported or explained again. MCP lets the AI tool retrieve eligible completed Wynter research as read-only context instead.
Wynter supplies the customer evidence. Your AI tool can use that evidence to analyze, compare, summarize or draft without you manually carrying the study into every conversation.
Use OAuth in ChatGPT or Claude, or a personal MCP token in Claude Code or Codex.
Connect through OAuth from the custom connector flow and approve access in Wynter.
Create a personal token in Wynter and add the Wynter MCP server to your AI client.
Use eligible completed surveys, message tests and preference tests without reconnecting.

Wynter’s Company data area includes the current setup instructions for connecting ChatGPT or Claude through OAuth.
The connector can list and read eligible completed surveys, message tests and preference tests. Message tests include full page, landing page and pricing page tests.
Tests using a custom audience and freemium tests are not exposed through MCP. Existing connections pick up newly supported test types automatically.
MCP gives connected AI tools access to research context; it does not let them edit or launch tests. Contact details typed by respondents are removed before results are served. Very large studies are returned in explicit parts instead of being silently cut.
Open Company data in Wynter and choose the connection that matches how you work. ChatGPT and Claude use the OAuth connector flow. Claude Code and Codex use a personal Wynter MCP token. The setup page shows the current MCP server URL and the exact steps for each client.
Wynter remains the source of the research. MCP changes where you can work with the evidence after a study is complete. Use the Wynter results experience when you want the original study view; use MCP when you want that evidence inside another AI workflow.
Connect your AI tool, ask it to list your completed tests, then bring the relevant study into the task you are already doing: analyze themes, compare evidence, prepare a brief or draft with the customer voice in context.