Connect your AI assistant to ResponseHub
The ResponseHub MCP server lets Claude, ChatGPT, Cursor and Claude Code work directly with your knowledge base, questionnaires and sources — with your permissions and full audit trail intact.
One server, whichever assistant you use
MCP is the standard way an AI assistant connects to an outside system. ResponseHub runs one MCP server that every assistant connects to — copy the URL from API → MCP in your account.
Web assistants
Add ResponseHub as a custom connector in Claude.ai or ChatGPT. You sign in and approve access — there is no token to manage. Best for asking questions and working with your knowledge base in conversation.
Developer tools
Connect Claude Code, Cursor or the Claude API using an API token passed as a header. Best for teams who want an assistant working alongside their own tooling and scripts.
https://app.responsehub.io/mcpEverything you do in ResponseHub, now in the chat
The MCP server exposes 21 tools across questions, your knowledge base, questionnaires, RFPs and sources. The assistant decides which one to use as you talk to it.
Ask questions, get cited answers
Run any question through ResponseHub’s full answer pipeline against your knowledge base and sources. Answers come back with a confidence score and citations to the exact source text.
Manage your knowledge base
Semantic search across approved items, then create, update, approve or remove entries — all without leaving the assistant. Answers can be auto-generated from your sources.
Run questionnaires and RFPs
List, create, download and delete questionnaires and RFPs. Start one from an XLSX or DOCX file and steer the answers with background context, straight from a conversation.
Handle source documents
Upload policies and other documents as sources, check when they are ready to answer from, and remove ones you no longer use — so the assistant is always working from current material.
Connected in a few minutes
Copy your server URL
Open API from your account menu, then the MCP tab, and copy your MCP server URL. It is the single endpoint every assistant connects to.
Connect your assistant
Add ResponseHub as a custom connector in Claude.ai or ChatGPT, or pass an API token as a header from Claude Code, Cursor or the Claude API.
Approve access
The assistant connects with your existing account permissions. It can only do what your role allows, and everything it does is recorded in your audit trail under your name.
Work in plain language
Ask “what does our policy say about data retention?” or “start a questionnaire from this spreadsheet”. The assistant picks the right tool from the 21 available and does the work.
{
"mcpServers": {
"responsehub": {
"url": "https://app.responsehub.io/mcp"
}
}
}Your controls come with it
Inherits your permissions
A connected assistant can never exceed what your account role permits. A read-only account gives you a read-only assistant — nothing more.
Fully audited
Every change an assistant makes appears in your audit trail under your name, so you always have a clear record of what happened and who triggered it.
Your data is never trained on
Data accessed through MCP is never used for training or fine tuning. AI models are accessed via AWS Bedrock, which operates with zero data retention.
See how the MCP server works under the hood
The help center walks through connecting Claude and ChatGPT, connecting developer tools, and the full reference for all 21 tools.
MCP server or JSON API — which should you use?
Both give an AI or a program controlled access to ResponseHub. Here is how they compare with pasting documents into a generic chat.
Want full control and automation? Use the API
The ResponseHub JSON API gives your own software programmatic control over knowledge base items, questionnaires, RFPs and more — ideal for automation you build yourself.
Bring ResponseHub into your AI assistant
Get started in under 5 minutes with our self-serve trial or contact us for a demo.
Frequently asked questions
What is the MCP server?
MCP (Model Context Protocol) is the standard way an AI assistant connects to an outside system. The ResponseHub MCP server lets assistants like Claude and ChatGPT work directly with your knowledge base, questionnaires, RFPs and source documents, so you can ask "what does our policy say about data retention?" and get an answer with a confidence score and citations, right inside the assistant you already use.
Which AI assistants can I connect?
Two kinds. Web assistants such as Claude.ai and ChatGPT connect by adding ResponseHub as a custom connector — you sign in and approve access, with no token to manage. Developer tools such as Claude Code, Cursor and the Claude API connect using an API token passed as a header. See the help docs for step-by-step setup.
Can an assistant do more than I can?
No. A connected assistant inherits your existing access controls and can never exceed what your account role permits. A read-only account gives you a read-only assistant, and every change an assistant makes appears in your audit trail under your name.
How is this different from the API?
The JSON API is for software you write — you decide exactly which endpoints to call and when. The MCP server is for assistants: it exposes the same capabilities as tools and lets the assistant work out which one to use during a natural conversation. Many teams use both — the API for automation they build, MCP for the work they do alongside an assistant.
Do you train AI on the data accessed through MCP?
Never. Your data is not used for any training or fine tuning. AI models are accessed via AWS Bedrock, which operates with zero data retention.
How do I connect it?
Copy the MCP server URL from inside ResponseHub — open API from your account menu, then the MCP tab — and add it to your assistant as a custom connector, or pass a token as a header for developer tools. Full instructions are in the MCP help docs.
Book a demo
Pick a time that works for you
Grab a 30-minute slot straight from the calendar. No forms, no waiting on a reply.
Book a demo