Seven companies, different clinical roles, and an Arizona connection.
An extra set of ears during an appointment. Help searching medical research. An alert that gets a concerning scan in front of a specialist.
“Healthcare AI” covers all of these. Understanding the differences makes the field much more interesting—and easier to evaluate.
Here’s a tour of seven companies working across individual practices, large health systems and specialized clinical settings. This is an educational snapshot, not a ranking or endorsement.
Smaller practices: help with the everyday workload
Heidi Health: less typing during the visit. Heidi turns clinical conversations into draft notes, with customizable templates for different specialties. Its published customer stories include solo practitioners and practice owners. The appeal is straightforward: more attention on the patient and less on the keyboard. Clinicians still need to review what the software produces. Its customer stories illustrate use, rather than independently establish results.
OpenEvidence: help with the “let me look that up” moment. OpenEvidence answers clinical questions using medical literature and provides citations. With individual clinician access and institutional offerings, it spans practice sizes. The useful distinction: finding evidence faster does not eliminate the need to check its quality or relevance to a particular patient.
Larger organizations: connecting AI to clinical workflows
Abridge: turning conversations into draft notes at scale. Duke Health announced an implementation intended to make Abridge available to 5,000 clinicians across more than 150 primary and specialty clinics. Clinicians review and edit generated notes before they enter the patient record. It illustrates the enterprise challenge: fitting documentation assistance into many specialties and locations.
Microsoft Dragon Copilot: dictation meets ambient listening. Dragon Copilot combines voice dictation, ambient documentation and AI assistance for clinical workflows. Microsoft has also introduced specialized nursing support. Available functionality depends on the organization’s implementation, making workflow fit an important part of any evaluation.
These are overlapping markets, not fixed divisions. A tool used by an individual clinician may also be deployed across a health system.
Beyond documentation: supporting detection and diagnosis
Viz.ai: connecting urgent findings with the right team. Viz.ai combines image analysis, alerts and care coordination. There’s an Arizona connection: Banner Health Foundation has reported its use at Banner–University Medical Center Phoenix and community hospitals across metro Phoenix. The platform helps providers share imaging and connect with stroke specialists around treatment and transfer decisions.
Aidoc: helping radiology teams flag urgent findings. Aidoc provides imaging AI and infrastructure for integrating it into clinical workflows. An announcement describing Advocate Health’s adoption reported an initial rollout across 22 sites, with algorithms flagging findings including pulmonary embolisms and intracranial hemorrhages. Adoption demonstrates real-world use; it does not establish that every implementation produces the same outcomes.
Digital Diagnostics: a specifically defined autonomous eye test. Its LumineticsCore system analyzes retinal images to detect more-than-mild diabetic retinopathy in eligible adults, without requiring a physician to interpret the images for that result. Its intended population includes adults 22 and older with diabetes and no previous diabetic retinopathy diagnosis. It brings a targeted diagnostic capability into settings such as primary care, but does not replace a comprehensive eye exam.
The question behind the impressive demo
Does the tool make care better, work easier, or both?
A published Sutter Health study of ambient documentation found less time spent on notes and lower reported mental workload. The change in burnout, however, was not statistically significant. Promising findings deserve attention—and careful interpretation.
For Arizona healthcare organizations, a useful starting point is to name the problem before choosing the product. Fewer unfinished notes, faster evidence searches and timely review of urgent scans call for different tools and different measures of success.
If your care team could hand one task to AI tomorrow, what would it be?
Sources and further reading
- Heidi’s customer stories
- OpenEvidence’s institutional offerings
- Duke Health’s Abridge announcement
- Microsoft’s introduction to Dragon Copilot
- Microsoft’s nursing workflow announcement
- Banner’s report on Viz.ai in Arizona
- Aidoc’s Advocate Health announcement
- LumineticsCore’s intended use
- Published ambient documentation study
October 2026 snapshot. Product capabilities and availability change. Company materials and adoption announcements are not comparative evidence of superiority.
