MCP for Claude, ChatGPT, and Co-pilot
Bring [cafeteria] into your existing workflows — context, sources, citations. Hear your consumers, verbatim.
Reading real conversations
cafeteria · query tables
answered from [cafeteria] 3,198 responses · 12 albums
Connect to your agent and get real consumer insights inside your chats. Brands, verbatim mentions, and demographics, all on demand.
01Find the connectors
Settings, then Connectors — web or desktop.
Settings›Connectors
02Hand over the URL
Add a custom connector and paste ours in.
https://mcp.cafeteria.page/mcp03Sign in and approve
Your work email, then the four read scopes.
Continue with [cafeteria]brandsmentionsdemographicsverbatims01Turn on Developer mode
Your profile › Settings › Security and login. It’s a setting inside ChatGPT, not your browser’s.
Settings›Security and loginDeveloper mode
02Add [cafeteria]
Plugins › +, or Settings › Apps › Create. Paste the whole URL, including /mcp, and keep auth on OAuth.
https://mcp.cafeteria.page/mcpOAuth03Sign in, then Create
Scan Tools first if it’s offered, sign in to [cafeteria], then save once the tools appear.
Continue with [cafeteria]Create01Open your agent’s tools
Copilot Studio, then add a tool.
Tools›Add a toolModel Context Protocol
02Hand over the URL
Name the server and paste ours in.
https://mcp.cafeteria.page/mcp03Sign in and publish
Your work account, then out to your tenant.
Continue with [cafeteria]Publish why not just chatgpt
Real people, real time, real citations. Here’s how [cafeteria] MCP stacks up.
[cafeteria] MCP
General AI
Traditional Research
Data source
Real Gen Next voices
Public web scrape
Recruited panels + fieldwork
Time to answer
Seconds
Seconds
4–8 weeks
Citations
Every claim linked to a real response
Often hallucinated
Report deck
Audience precision
Gen Z + Gen Alpha at scale
Whoever's online
Whoever gets recruited
Cost model
Flat subscription
Per token
Per study
ready to bring culture to your agent?
One server, and [cafeteria] answers inside the agent you already work in.