Dr. Mazin Shikara Gave His Doctors an AI Scribe. Here Is What Actually Changed, and What Didn’t.

The founder of MedFlorida Medical Centers ran the experiment. The results are more complicated than the vendors would have you believe.

TECHNOLOGY  ·  DR. MAZIN SHIKARA  ·  MEDFLORIDA MEDICAL CENTERS

The pitch for AI-powered clinical documentation is seductive in its simplicity. Doctors spend too many hours on electronic health records. An ambient scribe listens during the patient encounter, generates the note automatically, and gives physicians their time back. Problem solved. Next question.

The reality, as Dr. Mazin Shikara discovered after implementing Sunoh.ai across MedFlorida Medical Centers, is more instructive, and more honest, than that framing suggests. Some things improved meaningfully. Others did not move at all. Understanding the difference matters for every health system currently evaluating AI documentation tools, and there are a lot of them.

The Problem Was Never Just Time

Before explaining what changed, it is worth being precise about what the problem actually was. The conventional narrative frames physician documentation burden as a time management issue: doctors spend X hours per day on EHR tasks, and anything that reduces X is a victory. That framing is too simple.

Documentation burden is a cognitive load problem as much as a time problem. When a physician is mentally composing a SOAP note while a patient is describing symptoms, the physician is not fully present in that conversation. The diagnostic value of an unhurried patient encounter, the moment when a patient says something unrehearsed that changes the clinical picture, depends on the physician being genuinely attentive. A doctor who is simultaneously listening and formatting is doing neither well.

This is not a soft concern. Missed diagnostic cues, incomplete histories, and undertreated chronic conditions, these are the downstream consequences of a physician whose attention is structurally split. At MedFlorida, where the patient population skews heavily toward chronic disease, the consequence is visible directly: the patient whose blood pressure has been creeping upward for three visits but whose physician, rushing through documentation, did not flag the trend until the fourth.

“Technology should shorten the distance between a patient and an answer. When it lengthens that distance, it is a liability.”

DR. MAZIN SHIKARA

What Ambient Scribing Actually Does

Sunoh.ai operates passively during a clinical encounter. It listens to the physician-patient conversation, structures the content into clinical documentation format, and produces a draft note that the physician reviews and approves. The physician does not stop, dictate, or type mid-visit. The documentation happens in parallel with the encounter rather than consuming time after it.

The measurable effects at MedFlorida were real. Physicians completed their notes significantly faster. The carry-home documentation that many clinicians describe as the most demoralizing part of the job, finishing charts at 9 p.m. at a kitchen table, declined. End-of-day cognitive fatigue, as reported by staff, dropped in ways that were visible in clinical demeanor and patient interaction quality.

Patients noticed. Not because they were told about the technology, but because their physicians were looking at them during appointments instead of at screens. That shift is harder to quantify than a documentation time metric, but it is clinically real.

Accuracy was initially the primary concern. AI-generated clinical notes are only useful if they are reliable. In practice, the system required consistent physician review, which is appropriate and should be non-negotiable. The notes were not perfect; they required editing, particularly in complex multi-problem encounters. But the editing burden was substantially lower than composing from scratch, and review time decreased as the system learned encounter patterns.

What Did Not Change

Here is what the ambient scribe did not touch: the administrative burden generated outside the exam room.

Prior authorization requests. Insurance denials requiring physician-authored appeal letters. Referral paperwork. Prescription refill queues. Incoming lab result management. Coordination calls with specialists. These tasks consumed physician and staff time before the AI implementation and continued to do so after it. According to the American Medical Association, in some practices this non-documentation administrative layer represents 30 percent or more of total physician work hours. Ambient scribing does not reach that layer at all.

This is not a criticism of the technology. It is a calibration of expectation. Practices that implement AI scribing believing it will solve physician burnout comprehensively are likely to be disappointed. Practices that implement it as a targeted intervention for one specific friction point, the cognitive and time cost of real-time documentation, will see genuine returns.

The Workflow Lesson

The most important variable in AI scribing success is not the technology itself. It is the workflow it enters.

Ambient scribing deployed within an established longitudinal physician-patient relationship produces different results than the same tool used in a fragmented urgent care setting where the clinician has never seen the patient before. When a physician knows a patient’s history, the AI-generated note can be reviewed quickly because the physician already has the clinical context to evaluate it. When the physician is meeting a patient for the first time in a high-volume, high-turnover setting, review becomes more demanding.

MedFlorida’s model, same physician, repeated visits, integrated diagnostics, created conditions where ambient scribing worked well. That context is not incidental. A 2018 BMJ study linked higher continuity of care with lower mortality across multiple health systems, suggesting that the relational architecture of a practice shapes whether technology investments actually produce clinical value. Health systems evaluating AI scribing tools should assess not just the technology’s capabilities but whether their workflow architecture will allow those capabilities to manifest.

The Honest Verdict

AI-powered clinical documentation is not a solution to the physician workforce crisis. It is not a cure for burnout. It does not fix fee-for-service reimbursement, prior authorization abuse, or the administrative architecture that has turned medical practice into a clerical profession for many clinicians.

What it is: a genuine, meaningful reduction in one specific category of friction that compounds daily across millions of physician encounters. For primary care physicians managing chronic disease populations, where visit quality directly determines long-term patient outcomes, that reduction has real clinical value.

Dr. Mazin Shikara will continue using ambient AI scribing at MedFlorida. The evidence, from inside his own practice, supports it. But he would not describe it as transformation. He would describe it as buying back the afternoon, and in medicine, that is not nothing.

ABOUT THE AUTHOR

Dr. Mazin Shikara is the founder, President, and CEO of MedFlorida Medical Centers, a multi-location primary care network operating across Florida since 2007. He trained at the University of Baghdad College of Medicine, completed postgraduate training as a Member of the Royal College of Physicians in England, and finished his residency in internal medicine at Flushing Hospital Medical Center in New York. He is board-certified in internal medicine by the American Board of Internal Medicine.