The 12 research workflows teams actually run through Claude, Cursor and ChatGPT
Copy the prompt into your AI tool and run it before your next standup. Every recipe came out of what research teams do with their own data through the Great Question MCP, not out of a demo script.
Get all 12 recipes
An MCP use case for research is a job you hand to an AI agent that can reach your research data through the Model Context Protocol (MCP): finding the sessions that finished today, filing transcripts, pulling quotes across studies, checking whether a screener is converting. A research workflow recipe packages one of those jobs as a prompt, the list of Great Question MCP tools it calls, and the point where a person signs off.
The connection is what separates a recipe from a clever prompt. Without it, the model guesses. With it, the answer comes back with your own sessions, participants and transcripts attached, so you can click through and check it. These 12 recipes run in Claude, ChatGPT, Cursor and Copilot.
What's inside The jobs that eat a research week
Most of what teams ask an agent to do with their research data is retrieval and ops, not synthesis. The pack follows the usage.
Research ops
The daily interview digest, the transcript vault, evidence attached to the ticket, and bulk panel tagging.
Synthesis
Transcript to markdown, cross-study quote mining, and a highlight reel shortlist with links back to the source.
Recruiting
A response rate tracker, a panel feasibility check, and matching your CRM against the panel before you recruit.
Study design
Clone a screener that already works, and localise study emails without retyping them.
What we never automate
Four rules sit under every workflow here, and they matter more than the workflows. Each recipe names the tools it calls, so you can see what the agent touches before you let it near anything.
- Nothing reaches a participant without a person
- Nothing launches a study without a person
- Nothing reaches a stakeholder unchecked
- Nothing sees more than the person who connected it
MCP use cases: common questions
What are the most common MCP use cases for research teams?
Most are retrieval and ops rather than synthesis: a daily digest of completed interviews, filing transcripts where the team reads them, attaching evidence to tickets, tracking screener response rates and checking whether a panel can fill a study. Synthesis use cases, such as mining quotes across studies or shortlisting clips for a highlight reel, come next.
What is a research workflow recipe?
A prompt plus a live connection to your research data through the Great Question MCP. Each recipe names the tools it calls, who usually runs it, where a person signs off, and whether it is worth scheduling.
Which AI tools do the recipes work in?
Claude, ChatGPT, Cursor and Copilot. They all connect to the same Great Question MCP server, so a recipe works the same way in each.
Can an MCP workflow contact participants or launch a study on its own?
Not in this pack. No recipe sends an invitation, an email or a reminder, and none launches a study. The MCP can do those things; they are left out on purpose, so a person always presses go.
Which MCP workflows are worth running on a schedule?
3 of the 12: the daily interview digest, the transcript vault and the response rate tracker. The rest are better run by hand, when you need them.
Do I need a Great Question account?
Yes. The recipes read and write your Great Question data through the MCP, so you connect with your own account and see only what you can see in the app. Your admin decides whether the MCP is available. The MCP guide covers setup, and each recipe also has its own page, starting with the daily interview digest.
Connect first
The recipes run in any AI tool connected to the Great Question MCP.
Get the blog in your inbox
Latest insights for customer-obsessed product builders. Once a month, no noise.