We use Google Analytics and a Meta advertising pixel on our public site only — never on the clinical application. You can opt out at any time via “Cookie settings” in the footer. See our privacy policy.

All features
For AI agents & assistants

Works inside your AI agent.

CardioBridge exposes first-party tools to AI agents through a built-in MCP integration, and a public REST API alongside it. So heart-monitoring data — an ECG, a wearable trace — can be sent to our consultant cardiologists for review from directly within an agent workflow.

Built in

A first-party MCP integration.

CardioBridge ships an MCP / WebMCP integration that advertises a set of first-party tools to AI agents. Public tools are always available; a small set of account tools appears when the user is signed in — and even those return only non-identifying pipeline status.

This is real and in the product today. It is how an agent gets a trustworthy view of CardioBridge instead of guessing from a web page.

Discoveryjs
// CardioBridge advertises first-party tools to MCP-compatible agents via
// navigator.modelContext (WebMCP), with a window discovery fallback.
window.__cardiobridgeMcp.tools.map(t => t.name)
// [
//   "cardiobridge_get_service_overview",
//   "cardiobridge_get_api_overview",
//   "cardiobridge_submit_enquiry",
//   "cardiobridge_whoami",                 // signed-in only
//   "cardiobridge_list_my_ecg_statuses"    // non-identifying status only
// ]
Tools are advertised via navigator.modelContext, with a window fallback for custom agents.
How it flows

From agent to cardiologist.

An agent gathers an ECG or wearable trace, submits it to CardioBridge through the API, and the job enters the same worklist a clinic would use. A GMC-registered consultant cardiologist reviews the trace and signs a report, and the outcome returns to the agent workflow.

The agent never issues a diagnosis. It moves data and surfaces the result — every report is read and signed by a named human cardiologist.

What agents can do

Available today.

The tools an agent can call right now, all non-identifying:

  • Understand the serviceAgents can pull a structured overview of what CardioBridge does, who reads ECGs, turnaround and governance — straight into their reasoning.
  • Learn how to integrateAn API-overview tool returns the REST ingest flow and links to the docs, so an agent can wire up a submission path.
  • Check job statusFor a signed-in user, agents can list that user's own ECG jobs by reference and pipeline status — never patient names or clinical content.
  • Stay inside the guardrailsEvery agent tool runs through the same EU edge functions and Row-Level Security as the app, and exposes zero patient-identifying data.
What you can build

Bring your own agent.

The integration is agent-agnostic. CardioBridge does not have an official partnership with any specific assistant vendor — instead it speaks the open standards agents already use, so you can connect the agent you already build with.

  • MCP-compatible assistantsAny agent that speaks MCP — including assistants built on Claude or Gemini — can discover and call the first-party tools.
  • Custom agent workflowsYour own agents can call the public REST API directly to submit an ECG and receive the signed outcome via webhook.

Put a cardiologist in the loop.

Read the API docs to wire an agent to CardioBridge, or talk to us about your agent use case.

Read the API docs Talk to a specialist