Platform · MCP Server
Connect Claude, ChatGPT, and Any AI Agent to Your Practice Data
PatientCopilot ships with a Model Context Protocol (MCP) server. Your favorite AI tools can read your appointment data, query contacts, and trigger actions — through a secure standardized protocol.
What MCP is, in one paragraph
MCP — Model Context Protocol — is an open standard from Anthropic for connecting AI assistants to external data and tools. Instead of copy-pasting context into ChatGPT or Claude, the AI client speaks directly to a server that exposes structured tools. PatientCopilot runs one of those servers. The result: your AI tools can natively query and act on your practice data.
Why this matters for your practice
You are already using AI tools — Claude, ChatGPT, Gemini, Cursor — to draft, plan, and analyze. Until now, getting your practice data into those tools meant exporting CSVs, copy-pasting summaries, or building a custom integration. The MCP server replaces all of that with a one-time connection.
Real workflows it enables
"What does next week look like?"
Ask Claude or ChatGPT about your upcoming schedule, no-show risk, capacity gaps, and provider load — answered from live PatientCopilot data.
"Draft follow-ups for yesterday's consults"
Your AI assistant queries yesterday's appointments, pulls notes, and drafts personalized follow-up messages per patient.
"Which patients haven't booked in 90+ days?"
Get a segmented list with chief complaint, last-visit date, and suggested reactivation messages — without leaving Claude.
"Summarize this week's reviews"
AI client pulls every review across Google and Facebook, summarizes sentiment, and flags negatives for response.
n8n automations triggered from MCP queries
Build n8n flows that ask Claude (via MCP) for a decision, then trigger PatientCopilot actions based on the response.
Cursor + practice ops scripting
Developers and ops-savvy practice owners can write custom analyses in Cursor that query PatientCopilot live.
What's exposed
The current MCP server exposes 36 tools across the core PatientCopilot domains:
- Contacts — query, search, segment, and update patient records
- Appointments — read availability, list bookings, create / reschedule / cancel
- Conversations — read message history, send replies, route to humans
- Calendars — query provider schedules and capacity
- Opportunities & Pipelines — read deal flow, update stages
- Reviews & Reputation — pull recent reviews, draft responses
- Workflows — trigger automation flows
New tools are added continuously based on what practices and agencies actually use.
Compatible clients
- Claude Desktop + Claude Code
- ChatGPT (via custom GPT or Enterprise)
- Cursor (for AI-assisted development against your data)
- n8n via the n8n MCP node
- Any standards-compliant MCP client — the protocol is open
Security model
The MCP server uses scoped access tokens — you control exactly which data and which actions are exposed per token. Every tool call is logged for audit. For PHI workflows, confirm the AI client you connect (Claude Enterprise, ChatGPT Enterprise, etc.) has a BAA in place before sending PHI through it. See our post on whether AI software is HIPAA compliant for the broader picture.
How it pairs with the rest of PatientCopilot
The MCP server is your data layer for external AI. Agent Studio is the builder for internal AI. Knowledge Bases are what trains the bots. Ask AI is the in-platform assistant that uses the same data the MCP server exposes.
FAQ
What is MCP again?
Model Context Protocol — an open standard for AI clients to talk to external servers. Anthropic released it; major AI vendors now support it.
Is the MCP server included in the price?
Yes — included on all paid plans. No separate fee.
What if I have a custom tool not in the 36?
Use Agent Studio to build it, then expose the agent via MCP. Or request the tool — new ones ship frequently.
Does the AI client see all my data?
Only what your access token grants. Tokens are scoped per-tool and per-action; you can issue read-only tokens, single-domain tokens, or fully-scoped admin tokens depending on use case.
Ready to Get Started?
Contact us today and take the first step. Free consultations available.