Ambient clinical documentation is one of the earliest and most widely adopted products of these AMC-industry collaborations. Since its launch in 2025, Microsoft’s Dragon Copilot, a software that uses AI to automatically generate clinical notes from patient-clinician conversations within the Epic electronic health record (EHR), has been adopted by numerous AMCs nationwide. Building on this technology, Advocate Health, whose academic core is Wake Forest University School of Medicine, partnered with Microsoft to pilot and co-develop Project Nursing in 2024, an ambient AI documentation system that automatically captures spoken nursing observations and populates nursing flowsheets in Epic. The platform was designed for nursing workflows; input from frontline nurses guided its testing and refinement before broader deployment.

Northwell Health, the clinical enterprise affiliated with the Zucker School of Medicine at Hofstra/Northwell, partnered with the health-tech firm Abridge in 2025 to deploy an enterprise-wide ambient AI platform across its 28 hospitals as part of a systemwide digital transformation. Beyond automating clinical documentation, the collaboration aims to streamline downstream workflows including coding, risk adjustment, prior authorization, and nursing documentation. Houston Methodist, a leading teaching hospital in Texas and home of the Houston campus of Weill Cornell Medicine, partnered with Ambience Healthcare in 2025 to co-develop one of the first ambient AI documentation platforms designed for emergency departments and inpatient care. The platform integrates with Epic to generate specialty-specific clinical documentation, with emergency physicians and hospitalists helping to refine the product before broader deployment.

An emerging area of co-development is patient-facing conversational AI. Beth Israel Deaconess Medical Center, a teaching hospital of Harvard Medical School, partnered with Google DeepMind in 2026 to prospectively evaluate AMIE (Articulate Medical Intelligence Explorer), a conversational diagnostic AI for previsit patient history taking. Physicians supervised patient interactions and assessed the system’s safety and clinical performance before integration into routine workflows. DeepMind has since introduced a broader “AI co-clinician” model of triadic care, in which patients, physicians, and AI collaborate throughout a patient’s care while physicians retain the ultimate clinical authority.

Hartford HealthCare, a teaching affiliate of the University of Connecticut School of Medicine, partnered with the health tech company K Health in 2026 to develop PatientGPT, an AI-powered patient assistant that integrates with the EHR to provide personalized health education, explain laboratory results in plain language, identify potential medication interactions, schedule appointments, and escalate patients to care when needed. Unlike general-purpose chatbots, PatientGPT operates within the health system’s clinical infrastructure, allowing clinicians to review conversation summaries and maintain oversight.

Many AMC-industry partnerships focus on the digital infrastructure necessary to support clinical AI deployment at scale. UNC Health partnered with Microsoft in 2026 to implement Microsoft Fabric, a centralized data platform that enables clinicians and researchers to develop and deploy AI tools while maintaining patient privacy. The platform supports applications including EASI (Extract, Analyze, Synthesize, and Integrate), reportedly reducing manual chart review time by approximately 50%, and SHIRE (Secure Health Informatics Research Environment), a secure environment for AI-enabled clinical research. Similarly, Intermountain Health, the principal teaching hospital of the University of Utah Spencer Fox Eccles School of Medicine, partnered with Microsoft and Arize AI in 2025 to develop a secure cloud platform for building, deploying, and monitoring clinical AI applications. The collaboration aimed to reduce AI model deployment times from months to weeks, reportedly saving approximately 40 hours of technical work per AI product each quarter.

Duke Health launched a partnership with Trase Systems in late 2025 to co-develop and evaluate agentic AI tools designed to automate administrative tasks, improve patient scheduling and care coordination, and support personalized cardiovascular care. Clinicians at the Duke Heart Center work directly with Trase Systems to develop and test AI applications before broader deployment.

Some partnerships even extend to building the underlying health care–specific foundation AI models themselves. Mayo Clinic partnered with Microsoft in 2026 to develop a frontier health care AI model using deidentified clinical data, with Mayo validating the model in clinical practice before Microsoft distributes it through Azure AI Foundry to other health systems.

Precision medicine has become a key area of AMC-industry collaboration, with partners co-developing AI-enabled genomic and diagnostic technologies. Northwestern Medicine and PathAI announced a multiyear collaboration in mid-2025 to deploy PathAI’s AISight platform, a cloud-based digital tool that allows pathologists to view, organize, manage, and analyze digitized microscope slides with built-in AI tools that assist in reviewing tissue samples. The partnership also entails the development of AI-powered diagnostic models for applications ranging from cancer detection to biomarker quantification, with a goal of improving diagnostic accuracy and patient outcomes. Mass General Brigham, the largest teaching hospital system affiliated with Harvard Medical School, has partnered with GE HealthCare in a 10-year collaboration to co-develop AI solutions for computed tomography and magnetic resonance imaging and radiology workstations. The University of Pittsburgh School of Medicine partnered with the small firm Vizzhy Inc in 2025 to co-develop GAINMED, a generative AI platform that integrates multiomic data to support precision medicine. The Cleveland Clinic launched a collaboration with Aspira Women’s Health in 2026 to develop AI-enabled diagnostics for early detection and risk assessment of gynecologic diseases such as ovarian cancer and endometriosis.

The growing breadth of AMC-technology company partnerships suggests that AMCs are becoming an increasingly central component of the clinical AI innovation pipeline.

AMCs are helping to shape products during development, with the potential to shorten the interval between AI development and clinical deployment. The resulting AI systems may also be better aligned with real-world clinical workflows, helping bridge the gap between technical performance and demonstrated clinical benefit. Further, these partnerships may reshape how evidence for clinical AI is generated. Rather than relying on retrospective datasets, co-development opens more opportunities for prospective and stepwise evaluation across the development timeline.

However, tighter collaboration between AMCs and technology firms will require new bridges to be crossed. Policies governing the use of deidentified patient data, intellectual property ownership, revenue sharing, conflicts of interest, and ongoing monitoring after AI development will become progressively salient. Ultimately, the AMCs that harness tech industry co-development partnerships will be poised to shape the future of clinical care.