Accuracy without compromise: a programming-inspired solution for EHR workflow redesign in the generative AI Era

Ambient AI listening technology has emerged as a major advance in clinical documentation, reducing EHR burden and allowing clinicians to focus more fully on their patients. By pairing passive listening, speech-to-text transcription, and large language models, ambient systems can generate notes nearly in real-time and capture details often lost in traditional documentation. While these tools substantially reduce after-hours “pajama time” and support clinician well-being, they are not a complete fix. Documentation errors, propagation of inaccuracies, and persistent challenges with information review and retrieval remain. To address these gaps, we propose a comprehensive, programming-inspired approach to EHR workflow redesign modeled after the Language Server Protocol used in modern software development. The proposed standards-based platform would provide a shared infrastructure to integrate multiple generative AI and decision-support tools directly into clinical workflows. Core features would include context-aware autocompletion, inline error detection, hover information boxes, and structured chart review, transforming documentation from static recordkeeping into a dynamic, interactive process. By separating user interfaces from back-end services, innovations could scale rapidly across systems without each tool reinventing common functionality. This architecture aligns with broader efforts such as SMART on FHIR by emphasizing modularity, interoperability, and open standards. It promises to streamline information management, reduce redundancy and error, and consolidate critical data for timely decision-making. Future work will extend this approach beyond early use cases and evaluate its technical and methodological robustness through empirical studies in simulated and clinical settings.