What You Should Know

• The University of Pennsylvania Health System (UPHS) has announced a multi-year collaboration with primary care AI leader K Health to deploy a suite of integrated clinical agents across its network EHR platforms.
• The rollout kicks off within the Penn Medicine On-Demand virtual urgent care program before expanding into in-person primary care, cardiology, and dermatology clinics.
• The AI engine features a clinically validated, peer-reviewed intake interface that dynamically interviews patients to generate a pre-populated draft chart inside the provider’s EHR before the encounter begins.
• Built to eliminate point-of-care bottlenecks, the platform understands complex medical language, symptom profiles, medication tracking, and the clinical ambiguity native to everyday conditions.
• Backed by $384M in total venture financing, K Health and Penn Medicine will also co-develop peer-reviewed research to expand the global evidence base for routine clinical AI automation.

Beyond the Point Solution: Why Penn Medicine Is Integrating K Health AI Agents Across Its Core EHR

The front doors of America’s premier academic health systems are facing an unprecedented capacity crunch. Patient volumes are rising across both virtual and brick-and-mortar care channels, leaving clinical teams to combat severe administrative overhead and charting bottlenecks. Historically, provider organizations have attempted to manage this stress by layering disconnected digital health point solutions onto their existing infrastructure—deploying standalone chatbots, separate intake forms, and isolated text-based tools.

In practice, these fragmented applications often exacerbate data silos, forcing clinicians to manually transfer patient details and toggle between multiple screens. The resulting friction increases wait times, heightens the risk of information gaps, and pulls doctors away from the high-acuity decisions that demand their advanced medical training.

To dismantle these administrative silos and hardwire an automated, system-wide intake layer directly into its core infrastructure, the University of Pennsylvania Health System (UPHS) has finalized a multi-year partnership with K Health. Moving past superficial point solutions, Penn Medicine is integrating K Health’s clinical AI agents directly into its existing electronic health record (EHR) systems and public-facing portal interfaces. This deployment aims to streamline the pre-visit data loop, insulate physicians from charting fatigue, and establish a safe, navigable digital environment for patients throughout their care journey.

Pre-Populating the Chart: Automated Intake with Medical Context

The collaboration will initially launch within Penn Medicine On-Demand—the health system’s virtual urgent care framework—before expanding into the network’s in-person primary care clinics and high-volume specialty divisions, including cardiology and dermatology.

The engine driving this clinical automation is K Health’s peer-reviewed, clinically validated patient intake interface. When a patient initiates a request for care, the platform opens an active, conversational dialogue. Rather than prompting the user with a static, rigid questionnaire, the agent guides the patient through a dynamically personalized encounter tailored to their described reason for the visit.

As the patient interacts with the interface, the underlying algorithms translate natural symptoms, active medications, and historical care timelines into structured medical data. This automated synthesis generates a pre-populated draft clinical chart that flows directly into the physician’s EHR, remaining fully visible and ready for review before the consultation ever begins.

Mitchell Schnall, MD, PhD, Senior Vice President for Data and Technology Solutions at Penn Medicine, noted that the health system views AI as a significant clinical opportunity to optimize direct patient care. By deploying this native system-level integration, Penn Medicine can thoroughly evaluate how advanced AI models scale across the entire spectrum of routine institutional care.

Training AI on Real-World Medical Interactions

A primary reason general-purpose artificial intelligence models routinely fail when deployed in an acute healthcare setting is their fundamental lack of domain-specific clinical nuance. Medical communication is rarely binary; it is filled with shorthand terminology, nuanced diagnostic codes, and intense situational ambiguity.

K Health addresses this safety hurdle by training its enterprise clinical agents exclusively on large lakes of real-world medical interactions and verified clinical datasets. This specific grounding ensures the platform natively understands the language of medicine, allowing it to safely interpret complex patient inputs and match them to standardized clinical pathways.

Ran Shaul, Co-founder and Chief Product Officer of K Health, emphasized that as health systems globally race to construct their consumer-facing digital strategies, Penn Medicine is treating clinical AI as foundational infrastructure. The software functions as a continuous layer that prepares the provider visit, optimizes workflow coordination, and seamlessly routes patient inquiries into an auditable pathway.

To further cement this framework, Penn Medicine and K Health are aligning to launch collaborative, peer-reviewed clinical research initiatives. Building on K Health’s prior academic publications tracking chronic disease management and primary care automation, the joint research projects will focus on developing a rigorous, evidence-based benchmark for how autonomous agents impact workflow efficiency and patient compliance over time.