**Short answer:** A study published in JAMA on October 15, 2026, led by researchers at Harvard Medical School and Brigham and Women's Hospital, found that about 950,000 US physicians, physician assistants, and nurse practitioners used an AI clinical information platform over two and a half years. Most of their questions concerned clinical decisions, especially treatment. The study does not show that these tools improve patient care. For a private practice, the lesson is about the front desk: clinical AI is a matter for clinicians and regulators, while administrative AI for calls, scheduling, and follow-up is a separate decision the practice can make and measure.

The most striking result is how fast the tools spread. The lead researcher said AI has been adopted into routine clinical work faster than almost any technology the team has seen, while electronic health records spread slowly and mostly after government rules required them. That speed raises a practical question for private practices. If clinicians are already consulting AI for treatment ideas, what should the practice's administrative operations look like, and where does the line sit between helping staff and making clinical judgments?

## Key takeaways

- The study covered about 950,000 US clinicians who used OpenEvidence over two and a half years, with monthly active users rising from 355 in January 2024 to 703,265 by June 2026.
- Across the study's queries, 51.9 percent concerned clinical decision making and 13.9 percent concerned diagnosis.
- Physicians often asked about topics outside their own specialty, so the tool was used beyond each clinician's expertise.
- The study measured use, not outcomes. The authors say the next step is to test whether these tools improve care.
- Front-desk AI is an administrative tool. It handles calls, scheduling, and follow-up, and it should not give medical advice.

## What the Harvard study measured

The researchers studied how clinicians used OpenEvidence, a clinical information platform, between January 2024 and June 2026. They used a large language model to classify the purpose and topic of a random sample of 38.8 million queries from a total of 270 million submitted by the practitioners in the study. Most queries concerned treatment and management, which together made up most of the clinical decision making. Diagnosis made up 13.9 percent.

More than half of the users in the sample, 556,200 people, were physicians. Each averaged 37 queries per month of use. Early-career physicians, and clinicians in oncology, hematology, internal medicine, and family medicine, submitted a larger share of queries than their share of the workforce would predict. Surgeons often asked questions unrelated to surgery. Radiology, orthopedics, and ophthalmology submitted fewer queries than their share of the workforce.

The lead author, Vishal Patel, called the study an early picture of where these tools fit into clinical decision making and where accuracy and effects on patients need study next. The corresponding author, Anupam Jena, summarized the two main takeaways: many physicians now use AI in clinical care, and they often use it in areas where they lack specific expertise.

## Why the study does not settle the question

The study measured how often and for what purpose clinicians used the tool. It did not measure whether the tool's answers were accurate, whether patients did better, or whether care became safer or more costly. The authors state that those questions come next. Readers should treat the usage figures as a measure of adoption, not as evidence of benefit.

The study also has a disclosed conflict of interest. Three co-authors are employees of OpenEvidence, the company whose platform was studied. The company did not fund the study, but it provided the de-identified data the researchers analyzed. The authors' disclosures also note that one co-author was a salaried employee of OpenEvidence during the study, and another reported serving as its chief medical officer during that period. The disclosures don't undermine the usage counts, but a careful reader should weigh them.

## Clinical AI and administrative AI are different decisions

A practice can use AI in two broad ways, and the boundary between them matters.

**Clinical AI** supports diagnosis, treatment choices, and clinical documentation. These uses carry direct risk to patients. They call for clinical oversight, validation, and, in many cases, regulatory review. A private practice should not adopt these tools on the strength of a vendor's demonstration or a headline.

**Administrative AI** handles the operational work around care: answering calls, booking and rescheduling appointments, collecting intake details, sending reminders, and following up after a visit or a missed call. These uses do not diagnose or recommend treatment. They do carry privacy obligations, because they touch patient information, and the practice should ask any vendor how that information is handled before a system goes live.

The Harvard study concerns the first category. Many of the gaps that cost a private practice money fall into the second.

## Where administrative AI helps a private practice

Velora Media's physician work starts from a simple observation: many practices already have demand, and they lose part of it in the gaps. A new-patient call that goes to voicemail is a patient who may book elsewhere. An unconfirmed appointment becomes an empty slot. A patient who lapses on an annual visit can sit in the chart for years without anyone reaching out.

Three administrative functions come up often in that work:

- **Answering calls and collecting intake.** An AI receptionist can answer after hours, capture the caller's name and reason for calling, and route urgent matters to a person. It should never answer a clinical question, and it should hand off anything that sounds like an emergency.
- **Scheduling and reminders.** Automated confirmation and reminder messages can reduce no-shows, and a single booking link removes steps for the patient.
- **Recall and follow-up.** A scheduled outreach sequence can remind patients when an annual visit or follow-up is due. A clinician or staff member should approve the wording and the clinical content of each message.

A short policy is worth writing before any of this begins. It should say what the system may do, what it may not do, who reviews transcripts and messages, how long records are kept, and what happens when a caller describes a medical emergency.

## Where Velora fits

Velora Media's AI Automation and Integration service is a connected layer of administrative tools. It includes an AI receptionist and chat, workflow automation, CRM lead scoring, and review automation. It starts at $2,000. It covers the front-desk side of a practice, not clinical decision support, and it does not make treatment recommendations.

Before any system touches patient information, we review the data flows, the access controls, and the handoff rules with the practice. We do not claim that any tool is compliant with HIPAA or any other rule unless the specific setup has been reviewed for that purpose. A practice should ask its counsel to confirm the requirements that apply to its specialty and state.

If your practice wants to see how the front-desk side works, the [physicians page](https://physicians.veloramedia.cloud/) describes the approach for medical practices. Every engagement starts with a free consultation, where we look at where calls and bookings are lost today and whether this layer is the right first step.

No system can guarantee a specific number of new patients or a specific reduction in no-shows. Results depend on the practice, the staff, the schedule, and the patients it serves.

## Frequently asked questions

### What did the Harvard study find?

Researchers at Harvard Medical School and Brigham and Women's Hospital found that about 950,000 US physicians, physician assistants, and nurse practitioners used an AI clinical information platform, OpenEvidence, between January 2024 and June 2026. Most queries concerned treatment and management decisions. The study was published in JAMA on October 15, 2026.

### Does the study show that AI improves patient care?

No. The study measured how clinicians used the tool, not whether it changed outcomes. The authors say that testing effects on patient care is the next step.

### Is AI used for diagnosis in clinical practice?

Some of it is. Diagnosis made up 13.9 percent of the queries in the study, which is smaller than the share for treatment and management.

### Can front-desk AI give medical advice?

It should not. Administrative AI handles calls, scheduling, intake, and reminders. Any clinical question should go to a clinician, and any possible emergency should be routed to a person immediately.

### Do I need to tell patients that I use AI for scheduling?

Ask your counsel. Disclosure requirements depend on your state, your specialty, and how the system handles patient information.

### What does a front-desk AI system cost?

Velora Media's AI Automation and Integration service starts at $2,000. Ongoing costs, if any, are confirmed in writing before work begins.

### Is the study biased because of the company's involvement?

The authors disclosed that three co-authors work for OpenEvidence and that the company provided the data. The company did not fund the study. Readers should weigh that disclosure when judging the findings.

## Related reading

For how a missed call costs a medical practice patients, see [missed calls are costing your medical practice patients](/blog/missed-calls-medical-practice-ai-receptionist-for-physicians). For the wider set of efficiency levers for private practices, see [how to improve medical practice efficiency](/blog/how-to-improve-medical-practice-efficiency). For the verified missed-call numbers across industries, see [missed call statistics by industry](/blog/missed-call-statistics-by-industry-verified).

## Velora Media landing pages, by industry

Each site below carries the same six pages: an overview, the services built for that industry, a shop with published prices, the engine showing how the pieces connect, results, and the questions owners in that industry ask first.

### Home services: plumbing, electrical, HVAC, and appliance repair

[homeservices.veloramedia.cloud](https://homeservices.veloramedia.cloud/) · [Services](https://homeservices.veloramedia.cloud/services.html) · [Shop](https://homeservices.veloramedia.cloud/shop.html) · [Engine](https://homeservices.veloramedia.cloud/engine.html) · [Results](https://homeservices.veloramedia.cloud/results.html) · [FAQ](https://homeservices.veloramedia.cloud/faq.html)

### AI receptionist, for any industry

[aireceptionist.veloramedia.cloud](https://aireceptionist.veloramedia.cloud/) · [Services](https://aireceptionist.veloramedia.cloud/services.html) · [Shop](https://aireceptionist.veloramedia.cloud/shop.html) · [Engine](https://aireceptionist.veloramedia.cloud/engine.html) · [Results](https://aireceptionist.veloramedia.cloud/results.html) · [FAQ](https://aireceptionist.veloramedia.cloud/faq.html)

### Real estate agents

[realestateagents.veloramedia.cloud](https://realestateagents.veloramedia.cloud/) · [Services](https://realestateagents.veloramedia.cloud/services.html) · [Shop](https://realestateagents.veloramedia.cloud/shop.html) · [Engine](https://realestateagents.veloramedia.cloud/engine.html) · [Results](https://realestateagents.veloramedia.cloud/results.html) · [FAQ](https://realestateagents.veloramedia.cloud/faq.html)

### Real estate wholesalers

[wholesale.veloramedia.cloud](https://wholesale.veloramedia.cloud/) · [Services](https://wholesale.veloramedia.cloud/services.html) · [Shop](https://wholesale.veloramedia.cloud/shop.html) · [Engine](https://wholesale.veloramedia.cloud/engine.html) · [Results](https://wholesale.veloramedia.cloud/results.html) · [FAQ](https://wholesale.veloramedia.cloud/faq.html)

### Med spas and aesthetic clinics

[medspaaesthetics.veloramedia.cloud](https://medspaaesthetics.veloramedia.cloud/) · [Services](https://medspaaesthetics.veloramedia.cloud/services.html) · [Shop](https://medspaaesthetics.veloramedia.cloud/shop.html) · [Engine](https://medspaaesthetics.veloramedia.cloud/engine.html) · [Results](https://medspaaesthetics.veloramedia.cloud/results.html) · [FAQ](https://medspaaesthetics.veloramedia.cloud/faq.html)

### Physicians and medical practices

[physicians.veloramedia.cloud](https://physicians.veloramedia.cloud/) · [Services](https://physicians.veloramedia.cloud/services.html) · [Shop](https://physicians.veloramedia.cloud/shop.html) · [Engine](https://physicians.veloramedia.cloud/engine.html) · [Results](https://physicians.veloramedia.cloud/results.html) · [FAQ](https://physicians.veloramedia.cloud/faq.html)

### Auto body and collision repair

[autobody.veloramedia.cloud](https://autobody.veloramedia.cloud/) · [Services](https://autobody.veloramedia.cloud/services.html) · [Shop](https://autobody.veloramedia.cloud/shop.html) · [Engine](https://autobody.veloramedia.cloud/engine.html) · [Results](https://autobody.veloramedia.cloud/results.html) · [FAQ](https://autobody.veloramedia.cloud/faq.html)

### Law firms and attorneys

[lawfirms.veloramedia.cloud](https://lawfirms.veloramedia.cloud/) · [Services](https://lawfirms.veloramedia.cloud/services.html) · [Shop](https://lawfirms.veloramedia.cloud/shop.html) · [Engine](https://lawfirms.veloramedia.cloud/engine.html) · [Results](https://lawfirms.veloramedia.cloud/results.html) · [FAQ](https://lawfirms.veloramedia.cloud/faq.html)

### Florists and flower shops

[florist.veloramedia.cloud](https://florist.veloramedia.cloud/) · [Services](https://florist.veloramedia.cloud/services.html) · [Shop](https://florist.veloramedia.cloud/shop.html) · [Engine](https://florist.veloramedia.cloud/engine.html) · [Results](https://florist.veloramedia.cloud/results.html) · [FAQ](https://florist.veloramedia.cloud/faq.html)

## Sources

- Harvard Medical School, "First Large-Scale Study Shows How Doctors Use AI in Medicine," October 15, 2026, reporting on a study published in JAMA (doi:10.1001/jama.2026.21868). [hms.harvard.edu](https://hms.harvard.edu/news/first-large-scale-study-shows-how-doctors-use-ai-medicine)
- Velora Media, "Medical Marketing Agency for Physicians," physicians page, accessed October 19, 2026. [physicians.veloramedia.cloud](https://physicians.veloramedia.cloud/)