Answers
An MCP server is a service that implements the Model Context Protocol so AI assistants like ChatGPT, Claude, or Perplexity can read and reason over external data. CorpusIQ runs a multi-tenant MCP server that exposes 25+ business tools. Each tool is read-only. Answers are cited. Setup is OAuth-only, with no service account credentials shared.
Related pillar: Model Context Protocol overview
The MCP server sits between the AI assistant and the business tools. The operator authorizes each business tool with read-only OAuth into the MCP server. When the assistant needs data, it calls the MCP server, which executes the read against the right tool and returns the result with stable source identifiers for citation.
Direct API access from an assistant to every tool would require the operator to share long-lived credentials and would not include a unified citation pattern. The MCP server centralizes OAuth, read-only enforcement, and citation, so the operator only manages one set of authorizations.
CorpusIQ operates a managed multi-tenant MCP server. Each operator's authorizations are isolated from every other operator. The server does not store the operator's files or records; it reads on demand and returns the result without persisting it.
ChatGPT (via remote MCP server support), Claude (native MCP support in Claude Desktop, Claude Web, and Claude Code), and Perplexity (remote MCP connection from the Perplexity interface) all connect to the same CorpusIQ MCP server on one subscription.
25+ read-only connectors. ChatGPT, Claude, and Perplexity. Solo $29.95 a month. 30-day free trial.
