Nelson Advisors: Epic Systems’ "Connectors" and the Emergence of Agentic Interoperability in Healthcare


Executive Summary
At its October 2026 Open@Epic executive gathering, Epic Systems formally announced "Connectors," an integration technology enabling autonomous and semi-autonomous artificial intelligence agents to search for data and execute operational actions across external, third party software environments. For nearly two decades, healthcare interoperability has concentrated on data liquidity, the passive, transactional exchange of clinical records facilitated by Health Level Seven (HL7) interfaces, Fast Healthcare Interoperability Resources (FHIR) APIs and federal exchange networks under the Trusted Exchange Framework and Common Agreement (TEFCA). Connectors represent an evolution from passive data retrieval to distributed task orchestration.
By linking Epic’s native artificial intelligence ecosystem, which encompasses the persona-based assistants Emmie, Art and Penny, alongside custom agents built via Epic's Agent Factory, with disparate enterprise systems, Connectors target the pervasive operational friction known as "swivel-chair" work. This transition addresses workflow fragmentation across acute operations, patient navigation, supply chain tracking, and revenue cycle management.
However, delegating transactional write-back capabilities to algorithmic agents across distributed software architectures introduces complex technical and policy challenges. The architecture relies on emerging standards such as the Model Context Protocol (MCP), introduces administrative gating mechanisms to prevent unauthorised data ingress or egress, and establishes a commercial marketplace model on the Epic Showroom that will fundamentally recalibrate health system IT strategies and vendor economics.
The Architectural Genesis of Connectors
The introduction of Connectors directly addresses structural deficiencies inherent to legacy healthcare software integration. In contemporary health systems, clinicians and administrative personnel frequently navigate a fragmented software landscape, toggling between the core electronic health record (EHR), specialised patient engagement portals, transport dispatch platforms, and clearinghouse billing engines.
While the widespread adoption of SMART on FHIR containers over the past decade partially mitigated this fragmentation by embedding external web applications within the EHR interface, these implementations typically functioned as isolated visual displays. Clinicians were still forced to manually transcribe data points between the embedded window and the primary clinical chart, preserving human cognitive burden.
Connectors fundamentally alter this paradigm by transforming integration from a visual embedding into a direct agent-to-application tool execution model. Derek De Young, who leads Agent Factory development at Epic, explicitly positioned Connectors as the solution to this chronic "swivel-chair" burden, defining the technology as an automated bridge that prevents personnel from jumping between disconnected applications to complete interrelated tasks.
Johnston Thayer, Vice President of Health Informatics at Epic, contextualised this evolution through an automotive parallel, contrasting legacy predictive alerts and clinical co pilots with modern agentic frameworks. Where assistive AI functions like lane-assist technology, guiding a human who remains entirely in control of the mechanical apparatus—agentic AI functions like an autonomous vehicle.
Operating an autonomous agent requires infrastructure explicitly designed for machine interaction rather than human visual interaction; agents cannot be expected to achieve operational efficiency by clicking visual buttons or navigating human centric interfaces. By embedding standardised, programmatic tool-calling directly into Epic's core infrastructure, Connectors position the EHR not merely as a database of record, but as an active coordination engine across enterprise health IT.
Architectural Dimension | Legacy Health IT Integrations (HL7 v2 / Direct FHIR) | Agentic Connectors Architecture (MCP-Enabled) |
Primary Interaction Model | Deterministic, transactional data exchange (Push/Pull payloads). | Agentic context synthesis and autonomous tool-calling. |
Execution Trigger | Event-driven programmatic hooks (e.g., ADT events, order signing). | Natural language intent parsing, predictive agents, and clinical events. |
System Visibility | Rigid database schema mapping via middleware bridges. | Dynamic tool and resource discovery via standardized metadata schemas. |
Action Capability | Limited to predefined read-only or tightly scoped RESTful write endpoints. | Multi-step bidirectional execution spanning external databases and enterprise platforms. |
User Interface Footprint | External web applications presented in isolated SMART-on-FHIR modal windows. | Native execution within the primary clinical context, abstracting third-party interfaces. |
Technological Architecture, Protocol Foundations and Agent Integration
The underlying design of Connectors bridges natural language reasoning, structured schema definitions, and secure remote procedure execution. Rather than engineering bespoke, point-to-point APIs for every external enterprise application, Epic has aligned its Connector infrastructure with the Model Context Protocol (MCP), an open communication standard that standardises how language models discover and invoke external tools and resources.
Within this MCP-aligned framework, third-party software applications function as MCP servers, exposing two distinct operational primitives: Resources and Tools. Resources constitute passive, readable data structures—such as equipment availability databases, ambulatory arrival schedules, or insurance plan formularies. Tools represent executable functions that modify external application state, such as submitting a prior authorisation query, reserving an assistive medical device, or dispatching an electronic order to a dietary kitchen.
Epic’s orchestration engine operates as the centralized MCP client. When an agent evaluates an operational requirement, it queries the local registry of enabled Connectors, parses the parameters, constructs an authenticated JSON-RPC payload, and dispatches the execution instruction to the external system.
This protocol structure integrates directly with Epic’s native artificial intelligence offerings, each tailored to distinct operational cohorts. In the ambulatory and inpatient charting domain, the clinician assistant Art utilizes Connectors to query external diagnostic data repositories, check logistical schedules across affiliate clinics, and route clinical documentation directly into external registries without requiring the clinician to exit the primary chart. Within the MyChart patient portal, the patient-facing assistant Emmie leverages
Connectors to interpret patient conversational requests and execute actions across ancillary health system tools, such as indoor wayfinding and logistical tracking. In administrative and operational departments, the financial assistant Penny employs Connectors to communicate with clearinghouse portals and payer transaction platforms, automating claims adjudication and denial follow-up.
Beyond these out-of-the-box assistants, Connectors are engineered to plug directly into Agent Factory, Epic’s visual, low-code/no-code builder environment. Agent Factory enables health system analysts and enterprise IT architects to construct organisation-specific autonomous workflows. Analysts can assemble multi-step workflows that incorporate institutional clinical protocols, route approvals through local hierarchies, and link these internal logic trees directly to external Connectors, creating highly customised automation without building dedicated middleware software.
Cross Application Workflows and Clinical Operational Use Cases
Connectors operate across two functional paths: real-time contextual information retrieval and active task completion within external systems. By supporting multi-step tool execution across external platforms, Connectors support workflows that transcend traditional organisational silos.
In acute and emergency environments, clinical outcomes depend on rapid information processing. Connectors allow emergency departments to interface directly with external computer-aided dispatch (CAD) and regional ambulance telemetry software. En route to the hospital, real-time vehicle transit times and streaming patient metrics are ingested by Epic agents via a dedicated Connector. The agent continuously updates the trauma bay schedule, alerts on-call clinical specialists, and prepares admission documentation before the physical transport arrives, removing critical delays during patient handoff. Similarly, during complex inpatient surgical preparation, Art queries external durable medical equipment (DME) management platforms to confirm the inventory and readiness of specialised recovery devices, preventing post-operative discharge delays.
In patient facing settings, Connectors streamline complex logistics. Hospitalised patients conversing with Emmie through MyChart can order meals using natural language; the agent evaluates physician-ordered clinical dietary restrictions inside Epic and transmits the approved order directly to the hospital’s third-party food services software.
For outpatient care, Connectors integrate with specialized wayfinding platforms such as Gozio Health. Emmie coordinates appointment reminders with step-by-step navigational guidance, routing patients through parking facilities, clinical departments, diagnostic laboratories, and pharmacy collection points.
Furthermore, agents can chain multiple Connectors together to resolve complex logistical arrangements. In one modelled scenario, Emmie plans an appointment for an elderly patient lacking reliable transportation by simultaneously querying transit APIs, evaluating local weather forecasts, and adjusting the scheduling window to minimise safety risks.
Administrative and revenue cycle operations experience similar consolidation. Connectors link the Penny agent directly to national transaction networks, such as Availity, to automate coverage verification, prior authorization processing, and claims attachment distribution. Instead of staff manually logging into proprietary health plan portals to upload supporting records, Penny identifies required clinical criteria, packages relevant chart sections, and transmits requests directly through payer specific transaction rails.
In outpatient medication tracking, Connectors interface with prescription platforms like VPLRx, enabling care teams to view real-time shipping milestones and provide patients with verified arrival dates before discharge.
Operational Domain | Partner / Integration Target | Specific Functional Task Completed via Connector | Primary Persona / Agent |
Emergency Logistics | Regional CAD & Telemetry Systems | Ingests real-time transit telemetry and ambulance arrival predictions directly into ED trackboards. | Art (Clinician) |
Inpatient Support | Hospital Food & Dietary Management Systems | Validates physician diet orders and dispatches patient-requested meal orders into external kitchen prep systems. | Emmie (Patient) |
Patient Wayfinding | Gozio Health Platform | Generates integrated parking assignments, building navigation, and intra-facility routing within MyChart. | Emmie (Patient) |
Prescription Logistics | VPL / VPLRx Logistics Software | Pulls real-time tracking metrics to display confirmed medication delivery timelines for discharged patients. | Art / Emmie |
Revenue Cycle | Availity Integration Network | Executes prior-authorization eligibility queries, routes claims attachments, and verifies payer responses. | Penny (RCM) |
Enterprise IT Support | Enterprise Service Desk / Ticketing Platforms | Interrogates external service management logs to contextualize clinician technical support and configuration updates. | Agent Factory Agents |
Enterprise Governance, Security Architecture and Bounded Agency
Permitting autonomous AI agents to initiate transactions across enterprise boundaries introduces acute cybersecurity, clinical safety, and data privacy vulnerabilities. An unconstrained agent executing invalid write-backs or exfiltrating protected health information (PHI) poses substantial institutional liability. In response, Epic has engineered Connectors around three defensive tiers: mandatory administrative gating, aggressive adversarial stress-testing, and bounded algorithmic autonomy.
Administrative gating ensures that Connectors are not globally activated without institutional evaluation. The authority to enable a Connector resides entirely with the individual healthcare provider organization. Before any integration is brought online, hospital cybersecurity, compliance, and clinical governance teams must review an explicit permission manifest detailing what clinical resources the Connector can read, what state changes it can execute, and what user roles or patient cohorts are authorised to trigger it. Data pipelines cannot initiate transactions without this institutional clearance.
To safeguard this architecture against novel cyber threats and prompt injection vectors, Epic has utilized advanced frontier models, such as Anthropic’s Claude Mythos, to conduct defensive penetration testing against its medical-records infrastructure. These simulated red-team exercises identify misconfigurations in data pathways, verify boundary enforcement between multi-tenant environments, and ensure that dynamic agentic tool-calling cannot be hijacked to achieve unauthorized administrative privilege escalation. Epic pairs these adversarial tests with automated simulation batteries that run continuous evaluations against Emmie, ensuring conversational interactions remain within safe clinical boundaries.
Crucially, Epic enforces bounded agency rather than unrestricted autonomy. Johnston Thayer emphasised that Connectors expand the contextual data available to agents, but do not grant them unmonitored decision-making authority. Bounded autonomy requires that high-risk transactional workflows, such as clinical order confirmation, complex diagnostic routing, and legally binding financial appeals, retain mandatory Human-in-the-Loop (HITL) checkpoints.
Within Agent Factory, health systems can configure deterministic execution paths alongside probabilistic steps, ensuring agents escalate ambiguous contexts to human supervisors. Every action, tool invocation, and data payload passing through a Connector is comprehensively logged, creating an immutable audit trail for institutional compliance officers.

Commercial Distribution, Marketplace Economics and Implementation Trajectory
Epic’s rollout plan for Connectors integrates software development previews, collaborative customer build-a-thons, and centralised commercial distribution through the Epic Showroom.
Deployment Phase | Targeted Calendar Window | Strategic Milestone and Technical Scope | Operational Stakeholders |
Technical Incubation | May 2026 | Initial Agent Factory Build-a-Thon; collaborative development of early agentic prototypes. | Epic Core R&D and Early Adopter Health Systems. |
Early Adopter Training | October 2026 | Hands-on cohort training on agent configuration, governance tooling, and permission sandboxing. | Health System Informatics Teams & Certified Analysts. |
Partner Field Pilots | Late 2026 | Controlled testing of external Connectors across live inpatient and ambulatory environments. | Pilot Health Systems, Availity, Gozio Health, and VPL. |
Commercial Launch | March 2027 | Public distribution of developer-created Connectors via the Epic Showroom marketplace. | Global Epic Customer Base and Third-Party Software Developers. |
The commercial strategy underpinning Connectors marks a transition in Epic’s platform economics. Third-party software vendors wishing to distribute Connectors to healthcare provider organizations must publish them through the Epic Showroom marketplace. Epic will charge software developers listing and distribution fees to maintain the connection infrastructure, security sandboxing, and execution environments.
This monetisation structure establishes Epic as a gatekeeper and tollgate for healthcare operations. For third-party software vendors, listing a Connector becomes an operational imperative to ensure their tools can be invoked by Epic's agent layer, while Epic monetizes the underlying infrastructure that unifies clinical and operational computing.
Systemic Industry Implications and Strategic Outlook
The launch of Connectors introduces structural disruptions that extend far beyond technical integration, resetting competitive dynamics across the health technology ecosystem.
First, Connectors accelerate the centralization of the clinical and operational workspace within the EHR. Over the past decade, venture-backed digital health vendors successfully unbundled the EHR by deploying specialised point solutions for patient messaging, clinical documentation and scheduling. By enabling Emmie, Art, and Penny to execute actions directly within these specialised software applications, Epic effectively decouples the underlying service logic from the user interface.
The primary user interaction remains anchored within Epic’s interfaces, while third-party solutions are relegated to headless back-end utilities. Specialised health tech companies that rely on user engagement metrics, active portal logins, or branded digital interactions face disintermediation, as their core capabilities are ingested and presented directly through Epic’s agent layer.
Second, Connectors challenge the longevity of traditional Robotic Process Automation (RPA) and custom interface consulting. Healthcare organizations spend tens of millions of dollars annually maintaining screen-scraping bots and brittle interface engines designed to synchronize records between disconnected IT platforms. These legacy RPA workflows break frequently whenever an underlying application updates its user interface layout.
Because Connectors run on standardized, protocol-driven tool execution frameworks like MCP, they provide resilient, schema-validated execution channels that bypass graphical interfaces entirely. As health systems adopt native agentic integrations, demand for custom interface maintenance and brittle RPA bridges will rapidly decline.
Third, Connectors will reshape health data regulation and compliance. Federal policy under the 21st Century Cures Act and CMS interoperability frameworks has concentrated on preventing information blocking and establishing standardized, read-oriented FHIR APIs. As autonomous agents begin executing transactional tasks across software boundaries, regulatory focus must shift from data accessibility to action accountability.
When an AI agent misinterprets clinical context and triggers an improper write-back in an external clinical scheduling or dietary preparation app, liability frameworks must delineate fault across model developers, EHR platforms, integration protocol standards, and health system governance boards. Legal and risk management departments will face new demands to ensure agentic audit trails comply with evolving state and federal health IT mandates.
Strategic Conclusions
Epic Systems’ introduction of Connectors marks a pivotal evolution in enterprise healthcare technology. By building infrastructure that enables artificial intelligence agents to search for context and execute complex tasks across external enterprise software, Epic is addressing the systemic "swivel-chair" fragmentation that has burdened clinicians and administrators for decades.
Anchoring this integration in open, standardised specifications such as the Model Context Protocol provides an extensible framework for third-party developers, establishing an ecosystem far more resilient than legacy screen-scraping automations.
However, this technological leap cements Epic’s role as the central orchestration platform in healthcare, granting it significant operational and economic control over third-party digital health applications. Health system leaders must not view Connectors merely as convenience features; they represent a fundamental change in enterprise risk, clinical workflows, and software procurement.
Health systems must establish rigorous, multidisciplinary AI governance committees that bridge cybersecurity, medical informatics and legal compliance to review and audit every Connector permission manifest.
Moving toward the March 2027 marketplace launch, the healthcare organisations that realise the greatest operational gains will be those that implement clear human-in-the-loop safeguards, ensuring that autonomous agent execution remains transparent, safe, and aligned with organisational clinical standards.
Nelson Advisors > European Healthcare Technology Investment Banking
Nelson Advisors specialise in Mergers and Acquisitions for European HealthTech, MedTech, Digital Health, Healthcare IT, Healthcare AI companies in the Lower to Mid Market ranging from $25M to $250M EV. www.nelsonadvisors.co.uk
Healthcare.Digital is the Google News approved HealthTech and MedTech Thought Leadership platform for Nelson Advisors, positioning them as a specialised authority on European Healthcare Technology M&A and strategic corporate development. https://www.healthcare.digital
Nelson Advisors publish Europe's Leading Healthcare Technology Investment Banking Newsletter every week, join 5000+ HealthTech and MedTech subscribers today! https://lnkd.in/e5hTp_xb
Nelson Advisors LLP
Hale House, 76-78 Portland Place, Marylebone, London, W1B 1NT
Meet Nelson Advisors @ Events in 2026
Digital Health Rewired > March 2026 > Birmingham, UK
NHS ConfedExpo > June 2026 > Manchester, UK
HLTH Europe > June 2026, Amsterdam, Netherlands
HIMSS AI in Healthcare > July 2026, New York, USA
Bits & Pretzels > September 2026, Munich, Germany
HealthInvestor Healthcare Summit > September 2026, London, UK
World Health Summit 2026 > October 2026, Berlin, Germany
HLTH USA 2026 > October 2026, USA
Global Health Exhibition 2026 > October 2026, Riyadh, Saudi Arabia
Web Summit 2026 > November 2026, Lisbon, Portugal
MEDICA 2026 > November 2026, Düsseldorf, Germany
Leaders in Health Summit 2026 > November 2026, London, UK
Venture Capital World Summit > December 2026, Toronto, Canada
Meet Nelson Advisors @ Events in 2027
Digital Health Rewired > March 2027 > Birmingham, UK
Barclays Health Elevate > March 2027, London, UK
NHS ConfedExpo > June 2027 > Manchester, UK
HLTH Europe > June 2027, Amsterdam, Netherlands




































Comments