Why Electronic Medical Record Vendors Will Fail at AI (and what to do about it)

Electronic medical record (EMR) vendors, such as Epic and Cerner, dominate the healthcare information technology landscape, yet their approach to integrating artificial intelligence (AI) is poised for significant shortcomings. This failure stems from fundamental misalignments between vendor incentives and the needs of the end user, i.e. the medical provider. Nevertheless, opportunities exist for healthcare institutions to bring AI capabilities in-house creating their own tools and bypassing the vendor entirely. This not only unlocks the full power of AI upon the electronic medical chart, but gives health care institutions powerful leverage against deeply entrenched vendors. This talk will explain why current EMR vendors will fail at useful AI deployment, and then proposes a mechanism to avoid this outcome with in-house AI capabilities grounded in the idea of institutional "prompt control".
Brief Agenda:

• Introduction: Landscape of current AI tools in medicine. The unique needs of inpatient vs outpatient environments.
• EMR Vendor Challenges: How misaligned incentives lead to suboptimal AI deployment.
• Prompt Control: Reclaiming AI capabilities with in-house prompt engineering.
• Technical Blueprint: Deploying open-source AI in secure, clinician-centric systems.
• Q&A: Exploring benefits, challenges, addressing common objections