The Ballpark MCP (Model Context Protocol) connector lets you build and manage research studies directly from your AI clients like Claude, ChatGPT and Cursor. Describe your research goal in a prompt and get back a fully structured study, complete with questions and tasks, ready for you to review and publish.
We're in the process of connecting the entire research lifecycle to MCP, from drafting and recruiting through to analysis. The vision is to let you use Ballpark in a headless way, wherever you work, so your AI assistant becomes a full research partner, not just a shortcut to one part of the process.
The Ballpark MCP connector is in beta: Press the chat icon at the bottom right to speak to our team about getting access to the latest MCP features.
For the most in-depth technical details, including tool specs and authentication scopes, head to our developer docs.
What is the Ballpark MCP connector?
It's a remote MCP server that connects your AI assistant to Ballpark. Once it's connected, you can draft, refine, and publish a study without opening Ballpark at all.
Right now, you'll need to add Ballpark as a custom connector rather than picking it from a built-in list. As we add the remaining features to MCP, we'll be moving towards a more standard setup.
What you can do today
Create a study. Describe your research goal and get a live study built directly in your Ballpark workspace, ready to review and refine.
Edit a study. Refine wording, add or change questions, and adjust tasks using natural language.
Ask about a study. Get answers about a study's current questions, intro, outro, settings, or logic without leaving your assistant.
Coming soon
We're actively expanding what the connector can do. Here's what's next:
Audience selection and recruitment. Describe who you want to hear from and get matching audience options, so you can recruit directly from your AI assistant.
Submitting recruitment orders. Go from audience selection to placing an order without leaving your assistant.
Querying study data. Ask questions about your results and insights directly.
Searching existing studies. Find and reference past studies from your assistant.
Some of these may already be visible to early access workspaces. Press the chat icon at the bottom right if you'd like access sooner.
Setting it up
Open your AI assistant's settings for adding MCP servers. This varies by client, so check your assistant's documentation for where to add a remote MCP server.
Add Ballpark's MCP server using this URL:
https://edge.ballparkhq.com/api/mcpSave the connection. You'll be taken straight to Ballpark to authenticate your account through a secure OAuth flow. You never need to share a password with your assistant.
Once you're signed in, your assistant should confirm Ballpark is connected and you're ready to start drafting.
Using a company-managed AI account? Custom connectors are sometimes restricted to admins on tools like Claude for Work, so you may need to ask your workspace admin to add Ballpark for you.
Using it
Talk to your assistant the way you would a colleague:
"Draft a study to understand why users abandon checkout on mobile."
"Add a question about first impressions before the tasks."
"Remove the last question, it's too leading."
"Make the tasks more specific."
"Looks good, create the study."
Your assistant shows a live preview as it works. When you're ready to publish, just say so. If you're not signed in yet, your assistant will walk you through connecting your account first.