What is an AI-native EHR?

An AI feature can be added to an existing EHR to help with a particular task, such as drafting a note. That can be useful. It does not, by itself, establish how patient context, the inbox, and follow-up work together across the system.

Designing around AI from the data model up means considering how source information, the patient's history, and proposed actions stay connected. The review workflow matters as much as the generated text: a clinician needs to understand what supports a proposal and decide what happens next.

The term describes an approach to design, not a guarantee of accuracy or practice fit. Evaluate the source context, review boundary, and actual workflow rather than relying on the label.

Chart preparation starts with relevant history. Thyra's longitudinal approach brings the patient's timeline into documentation, so reviewing a draft can include more than the current encounter. The clinician still checks the information that supports the note.

Documentation becomes a draft to inspect. The Longitudinal AI Scribe prepares note content from patient context; the physician compares it with the sources and resolves missing or conflicting information before accepting it.

Protocol follow-through starts in the inbox. Thyra's Smart Inbox triages messages, results, and refills against practice protocols and prepares drafts for sign-off. The workflow connects the incoming task with context and a proposed action, while keeping the final clinical decision with the physician.

Thyra focuses on independent endocrinology and primary care. Its published workflows include the Smart Inbox, longitudinal documentation, and CGM review with source context and physician review.

Overlay runs alongside the EHR a practice already uses. Full EHR adds the chart when the practice is ready. That lets a team evaluate the workflow before making a separate decision about replacing its system of record. Any EHR connection is confirmed for the data source, permissions, and workflow before a pilot begins.

A useful evaluation includes both clinical work and data handling: what information is available, where proposals are reviewed, and who is responsible for the next step.