Your Patients Are Asking AI for Help.
Today, Your Website Can Answer.
WebMCP lets AI agents book appointments, find providers, and verify insurance on your healthcare website. Structured tool invocation. Zero PHI exposure.
No credit card required · Free
“Find me a cardiologist near downtown that accepts Aetna and has openings this week”
AI Agents Are Already Trying to Help Your Patients.
They're Failing.
Patients ask AI assistants for healthcare help. The AI screen-scrapes your website — guessing at form fields, parsing HTML, clicking buttons. It gets it wrong constantly. Forms fail. Patients give up. Your phone lines stay jammed.
Google released WebMCP — a proposed web standard co-authored with Microsoft. Instead of guessing, agents invoke structured tools your website publishes. Your scheduling form becomes a callable function with typed parameters.
Healthcare has a clear boundary between public data (directories, availability, insurance) and protected data (records, history, results). WebMCP tools only expose the public layer. Architecturally cleaner than any other industry.
The question isn't whether AI agents will interact with your healthcare website. They already are. The question is whether that interaction will be reliable, structured, and under your control.
From Scheduling Form to Agent-Callable Tool. In Minutes.
Two APIs: Declarative (HTML annotations) and Imperative (JavaScript). Both designed for progressive adoption.
If your scheduling page already has an HTML form, you're 90% done. Add a few attributes and your form becomes an agent-callable tool.
Before — Standard HTML
<form action="/schedule" method="POST">
<label for="specialty">Specialty</label>
<select name="specialty" required>
<option value="cardiology">Cardiology</option>
<option value="dermatology">Dermatology</option>
<option value="orthopedics">Orthopedics</option>
</select>
<label for="location">Preferred Location</label>
<input type="text" name="location" placeholder="ZIP or city">
<label for="date">Preferred Date</label>
<input type="date" name="date">
<button type="submit">Search Availability</button>
</form>After — WebMCP Annotated
<form action="/schedule" method="POST"
toolname="find_available_appointments"
tooldescription="Search for available appointment slots
by medical specialty, location, and preferred date.
Returns available providers with next open slots."
toolautosubmit>
<label for="specialty">Specialty</label>
<select name="specialty" required
toolparamdescription="Medical specialty such as
cardiology, dermatology, or orthopedics">
<option value="cardiology">Cardiology</option>
<option value="dermatology">Dermatology</option>
<option value="orthopedics">Orthopedics</option>
</select>
<label for="location">Preferred Location</label>
<input type="text" name="location"
placeholder="ZIP or city"
toolparamdescription="Patient's preferred location
as ZIP code, city name, or neighborhood.">
<label for="date">Preferred Date</label>
<input type="date" name="date"
toolparamdescription="Preferred appointment date
in YYYY-MM-DD format.">
<button type="submit">Search Availability</button>
</form>Key insight: You didn't write a single line of JavaScript. You added HTML attributes to a form that already existed. The browser computes a JSON Schema automatically — agents discover and call your tool with typed parameters.
Designed with Healthcare Privacy in Mind.
Public data tools. Clinical data stays isolated.
Tools only expose what you explicitly register. Everything else doesn't exist in the agent's world. Not because of a policy. Because of the architecture.
Patient → AI Agent
Structured tool calls (JSON)
WebMCP Tool Layer
Public data only
find_providercheck_insurancecheck_availabilityschedule_apptget_servicesget_locationsEHR / Clinical Systems
PHI stays here. Always.
- ✓Provider directory search
- ✓Appointment availability (open slots)
- ✓Insurance plan acceptance
- ✓Clinic locations and hours
- ✓Service descriptions
- ✓General contact information
Already public on your website. WebMCP makes it structured.
- ✕Protected Health Information (PHI)
- ✕Patient records or medical history
- ✕Test results, labs, imaging
- ✕Prescription information
- ✕Insurance claims or billing
- ✕Authenticated patient data
Not registered = doesn't exist for agents. By architecture.
Technical Enforcement
Explicit Registration
navigator.modelContextOnly contains tools you register. EHR, patient portal, clinical systems have no registration → invisible to agents.
Agent Detection
SubmitEvent.agentInvokedEvery agent-triggered form submission is flagged. Apply additional validation, logging, or restrictions.
Consent Gating
requestUserInteraction()State-modifying tools pause for patient confirmation. Architecturally enforced by the protocol.
Read-Only Annotations
readOnlyHint: trueInformation-retrieval tools signal they don’t modify data. Agents use this for smarter confirmation decisions.
9 Tools. Thousands of Patient Requests.
The complete healthcare WebMCP tool library. Public data only. Patient consent required for state-modifying actions.
find_providerSearch providers by specialty, location, insurance, language
get_provider_detailsFull bio, credentials, certifications, office locations
check_insuranceVerify plan acceptance for a provider or entire system
check_availabilityOpen slots by provider, date range, and visit type
schedule_appointmentBook appointment — requires patient confirmation
cancel_appointmentCancel by confirmation ID — requires patient confirmation
get_servicesService descriptions, departments, conditions treated
get_locationsAddresses, phone numbers, hours, directions
get_hoursOperating hours by location and department
4 tools. One seamless experience. Zero PHI exposure. Under 3 seconds total.
From Assessment to Agent-Ready. Tooling That Accelerates Each Step.
Implement WebMCP tools manually, or use our platform to generate tool definitions, test coverage, and validate security posture.
AI Co-Pilot
Your Virtual Implementation Team
Point our AI at your healthcare website. In minutes, it crawls every page, identifies all patient-facing forms, and generates HIPAA-aware WebMCP tool definitions for each one — with healthcare-specific best practices like requestUserInteraction() for scheduling and readOnlyHint for information retrieval.
- ✓Crawls every page and identifies all patient-facing forms
- ✓Generates HIPAA-aware WebMCP tool definitions
- ✓Flags over-parameterized forms collecting unnecessary patient data
- ✓Produces a prioritized implementation roadmap
- ✓Generates production-ready code for your tech stack
Designed to significantly reduce the time and effort required to define tools compared to manual implementation.
Healthcare Prompt Coverage Testing
200+ prompts across multiple AI models
We test your tools against 200+ healthcare-specific natural language prompts across Gemini, GPT, and Claude — from formal (“I'd like to schedule a dermatology appointment”) to casual (“need a skin doctor asap”) to multilingual (“Necesito un doctor que hable español”).
Security Scanner
Healthcare-specific vulnerability checks
Missing requestUserInteraction() for appointment booking.
→ Fix: Add requestUserInteraction() before POST call.
Parameter “patient_age” detected. Over-parameterization risk.
→ Fix: Remove patient_age or make non-required.
HIPAA Compliance Reports
Auto-generated documentation
- ✓Complete data inventory per tool
- ✓PHI exposure analysis verification
- ✓Consent mechanism audit
- ✓Agent interaction audit trail
- ✓Architecture diagram for compliance review
Your compliance officer reviews a pre-written document — not a blank page.
Agent Readiness Score
0-100 healthcare benchmark
Score benchmarks vary by healthcare vertical
WebMCP Adds a New Front Door. Your Backend Stays the Same.
WebMCP tools run client-side in the patient's browser. They call your existing backend APIs — the same APIs your website already uses. Your Epic, Cerner, or Athenahealth integration doesn't change. You're adding a new access channel, not rebuilding infrastructure.
Patient's Browser
Both call the SAME backend APIs
Your Backend (API Gateway)
Clinical Systems
UnchangedWhat's New
Just the WebMCP tool layer in the browser
What's Unchanged
Everything behind your API gateway
EHR-Specific Notes
Epic Integration
Your WebMCP check_availability tool calls your API gateway, which calls Epic's FHIR Scheduling API. The WebMCP tool never touches Epic directly. Your existing FHIR integration, security controls, and audit logging remain unchanged.
Cerner / Oracle Health
Same pattern — WebMCP tool → Your API → Cerner Millennium API. The tool layer adds structured agent access without modifying your existing Cerner integration.
Custom / Legacy Systems
If your scheduling runs on a legacy system with a custom API, WebMCP tools simply call that API. The declarative form approach is especially useful — annotate your existing forms and agent interactions route through the same submission path as human users.
Healthcare Organizations Preparing for AI Agent Interactions.
These are hypothetical scenarios illustrating how different healthcare organizations could use WebMCP. Results will vary based on implementation and adoption.
Regional Health System
Multiple locations · EHR-integrated · High call volume
Hundreds of daily phone calls from patients asking “who can I see near me for X?” Staff manually handle every one.
AI agents could answer the exact question driving those calls — provider search, insurance check, and availability — using structured tools on the health system’s public-facing website.
find_providercheck_insurancecheck_availabilityget_locationsVirtual-First Telehealth Platform
Online scheduling · Competing for visibility
Needs to be discoverable when patients ask AI assistants for telehealth options. Without structured tools, agents have no reliable way to surface the platform.
WebMCP tools would let agents discover available providers, check scheduling, and initiate bookings — giving the platform a structured presence in AI-assisted healthcare search.
find_providercheck_availabilityschedule_appointmentMulti-Specialty Practice
Small team · WordPress · No dedicated developers
Office manager wants the practice to be findable by AI assistants, but the team has no developer resources and limited technical budget.
Declarative HTML annotations on existing scheduling forms could make the practice agent-discoverable without writing code — the simplest path to agent readiness.
find_providercheck_availabilityget_servicesget_hoursThese scenarios are illustrative. Actual timelines and outcomes depend on your organization.
A Phased Approach to Agent Readiness. At Your Pace.
Implementation timelines vary by organization size and complexity. Here is a typical phased approach healthcare teams follow.
Audit & Assessment
Understand your current agent readiness and identify the highest-impact tools to implement first.
- ✓Run the Agent Readiness assessment on your website
- ✓AI Co-Pilot scans pages, forms, and provider directories
- ✓Receive a gap analysis: what agents can find vs. what they miss
- ✓Priority-ranked list of tools to implement
Tool Definition & Generation
The AI Co-Pilot generates your healthcare tool definitions. Your team reviews and customizes.
- ✓Co-Pilot auto-generates tool schemas from your site structure
- ✓Review each tool: name, description, parameters, return values
- ✓Customize descriptions for patient language
- ✓Configure consent flows for scheduling and data-sensitive actions
- ✓Set up rate limits, audit logging, allowed domains
Testing & Validation
Validate tools against real-world patient queries and review security posture.
- ✓Prompt Coverage Testing: healthcare query templates tested against your tools
- ✓Security Scanner validates consent flows and parameter safety
- ✓Architecture documentation generated for internal compliance review
- ✓Test tool descriptions across multiple AI models
- ✓Agent simulation: observe AI agents interacting with your tools
Launch & Optimize
Go live and iterate based on real usage data.
- ✓Deploy WebMCP tools to production
- ✓Monitor agent traffic in the dashboard
- ✓Track which tools agents use and which queries remain unresolved
- ✓Review competitive landscape as the ecosystem evolves
- ✓Iterate on descriptions and parameters based on real usage
Why Teams Choose WebMCP
Get your Agent Readiness Score in 60 seconds -- no signup required
Patients Are Already Asking AI for Help.
Prepare Your Healthcare Website to Answer.
More patients are turning to AI assistants for healthcare recommendations. With WebMCP, those assistants invoke your verified tools instead of guessing. Healthcare organizations preparing now will be ready when AI-assisted patient interactions become mainstream.
Enterprise Health System
Multi-location, Epic/Cerner integration. Talk to a specialist.
Talk to a Specialist60 seconds · No credit card · No signup required