Tool Directory

Physician AI Tools

AI tools physicians actually use in daily practice, grouped by what they do, with the regulatory status of each category stated plainly.

Start with the fact that governs everything else on this page: almost none of these tools are FDA-cleared medical devices, and that is by design rather than by oversight.

Software that drafts a note, summarizes a chart, or answers a reference question generally falls outside the device definition because a clinician reviews the output before it affects care. That exemption is what allowed this category to reach real adoption in three years while the cleared device list grew one submission at a time.

The tradeoff is that nobody reviewed these tools before they reached your clinic. There is no premarket evidence requirement, no standardized performance reporting, and no adverse event pathway. Evaluation is the buyer's job, and the buyer is frequently a health system committee rather than the physician who has to live with the output. See how to evaluate a clinical AI tool for the questions worth asking.

AI tools in physician workflows ... 21 records. Last updated 2026-09-09.
Tool Vendor Category Where It Runs Regulatory Status
Abridge Abridge Ambient Documentation Mobile and EHR-embedded Not a regulated device
Ambience Healthcare Ambience Healthcare Ambient Documentation EHR-embedded Not a regulated device
Augmedix Augmedix Ambient Documentation Mobile and service-based Not a regulated device
ClinicalKey AI Elsevier Clinical Reference and Answers Web and EHR-embedded Not a regulated device
CodaMetrix CodaMetrix Coding, Documentation Integrity, and Revenue EHR-integrated back end Not a regulated device
Consensus Consensus Literature and Research Web Not a regulated device
Elicit Elicit Literature and Research Web Not a regulated device
Glass Health Glass Health Clinical Reference and Answers Web Not a regulated device
In Basket Message Drafts Epic and Microsoft Inbox and Patient Communication EHR-embedded Not a regulated device
Microsoft Dragon Copilot Microsoft Ambient Documentation EHR-embedded Not a regulated device
Nabla Copilot Nabla Ambient Documentation Browser and EHR-embedded Not a regulated device
Nym Health Nym Health Coding, Documentation Integrity, and Revenue Back end integration Not a regulated device
OpenEvidence OpenEvidence Clinical Reference and Answers Web and mobile Not a regulated device
Oracle Health Clinical AI Agent Oracle Health Inbox and Patient Communication EHR-embedded Not a regulated device
PowerScribe Microsoft Imaging Reporting and Workflow Dictation platform Not a regulated device
Rad AI Rad AI Imaging Reporting and Workflow Dictation and PACS-integrated Not a regulated device
Regard Regard Coding, Documentation Integrity, and Revenue EHR-embedded Not a regulated device
SciSpace SciSpace Literature and Research Web Not a regulated device
Scite Scite Literature and Research Web and browser extension Not a regulated device
Suki Assistant Suki Ambient Documentation Mobile and EHR-embedded Not a regulated device
UpToDate Wolters Kluwer Clinical Reference and Answers Web, mobile, and EHR-embedded Not a regulated device

Why Ambient Documentation Won First

Documentation burden is the one problem in this category where physicians agreed on the diagnosis before anyone offered a treatment. Hours of after-hours charting is measurable, universally hated, and directly implicated in burnout. A tool that removes some of it does not need to prove clinical benefit to get adopted, which is a very different bar from the one an imaging device faces.

It also carries the lowest failure cost on this page. A badly drafted note is caught during review. A badly triaged stroke study is not.

The Category To Watch Most Carefully

Clinical reference tools are the ones where the regulatory exemption is doing the most work. A system that answers "what is the first-line treatment for this" is influencing a clinical decision, and the argument that a clinician's review makes it non-device rests entirely on that review actually happening. Under time pressure, review degrades into acceptance. That is not a hypothetical concern, it is the documented behavior pattern around every decision support system ever studied.

Ask any vendor in that category what the system does when it does not know, and whether it will say so.