Ambient Scribes: I Stopped Typing During Patient Visits. Here Is What Changed.

The Entrepreneur's Life

I Stopped Typing During Patient Visits. Here Is What Changed.

There is a specific body language that patients have learned to read in their physicians, and most of them have never named it out loud but they feel it immediately. It is the slight turn of the shoulders toward the screen. The eyes that drop to the keyboard. The half-present quality of a conversation happening with someone whose attention is divided between the person in the chair and the documentation field in the EHR. Most physicians who do this are not doing it because they stopped caring about the patient. They are doing it because the documentation requirements of modern medicine made the EHR the third party in every exam room conversation, and the patient learned to share their physician with a computer whether they wanted to or not.

I used an ambient AI scribe for the first time about two years ago. I want to tell you what that experience was like, why I have not gone back, and why I think every physician who still types during patient visits should take this technology seriously.

The Interim Solution That Worked for Nearly a Decade

Before the ambient AI scribe, I had a solution that worked remarkably well for almost ten years. It was not technology. It was people. I ran a nurse scribe model in my clinic in Plymouth, and at peak it took three nurses to support my workflow. That number tells you something about the scale of the problem documentation had become, and also something about how seriously I took solving it.

The nurse scribe model did things that ambient AI scribes still cannot do. My nurse scribes did not just create the note for me to review and sign. They executed verbal orders in real time during the visit. A medication order, a lab requisition, a referral, an immunization: these happened as the conversation produced them, managed by a skilled nurse who understood the clinical context and could act on it immediately. The patient experience of that visit was genuinely different from anything I have seen replicated in a purely technology-driven model. The nurse was a second clinical presence in the room, not just a documentation resource. She caught things. She flagged things. She was a human being who participated in the care of the patient in a way that a microphone and a language model do not.

I want to be honest about that, because I think the ambient AI scribe conversation sometimes moves too fast past what the human scribe actually was. It was not just a more expensive version of the same thing. It was a different thing that happened to solve the same documentation problem. For the physician who has the volume, the patient population, and the practice model to support it, the nurse scribe model remains one of the best clinical workflows I have ever seen. I built my practice around it for a reason.

But it is also expensive in a way that most practices cannot sustain indefinitely, and that became more true over time. Three nurses dedicated to scribing is significant overhead and headcount. That cost structure works in a high-volume independent practice with the right margin. It works less well in a locums context, in a hospital-employed setting, in a practice that is building toward a leaner model, or in any arrangement where headcount decisions belong to an institution rather than the physician. When I transitioned to locums work over the past few years, the nurse scribe model simply did not come with the job. What did come with the job was an ambient AI scribe already integrated into the clinical workflow. The comparison I could then make was direct, lived, and based on the same patient volume, the same documentation requirements, and the same physician doing both. The AI scribe won on cost and efficiency by a margin that was not close.

What it has not replaced, and I want to name this clearly, is the verbal order execution that made the nurse scribe model so clinically efficient. The ambient AI scribe listens, documents, and drafts. It does not act. It does not send the lab order or update the medication list or schedule the follow-up appointment as the clinical conversation produces those needs. That gap is real, and it is the reason I still think the nurse scribe model is worth considering for the physician whose practice can support the overhead. For everyone else, and for the settings where that overhead is not available, the ambient AI scribe is the most practical path to the same underlying goal: getting the documentation out of the way so the physician can be present in the room.

What the Screen Was Costing Me

I have practiced family medicine for more than thirty years. I know how to conduct a clinical conversation. I know how to listen for the thing behind the presenting complaint, the detail the patient mentions in passing that turns out to be the most diagnostically relevant piece of information in the encounter. I know how to read the nonverbal signals that tell me someone is frightened, or minimizing, or not telling me the whole story yet. These are skills built across decades of clinical practice, and they require one thing above all else: my full attention on the person in front of me.

The EHR took that. Not suddenly, but gradually, in the way that any new burden accumulates before you notice the weight of it. By the time I stepped back and looked honestly at my clinic workflow, I was spending a meaningful portion of every patient encounter with my eyes on the screen, my fingers on the keyboard, building the documentation in real time because the alternative was staying two hours after clinic to finish it. I had made a choice between being present in the room and being efficient with my time, and the documentation load had forced me into that choice without ever asking my permission.

The ambient scribe changed the terms of that choice. It did not eliminate documentation. Documentation still happens. But it moved documentation from something I was doing during the patient encounter to something that was being generated from the encounter, in the background, while I gave my patient my full attention. The difference in the quality of that attention is not subtle. My patients have noticed it. I have noticed it. And the clinical consequences of that restored attention are real in ways I did not fully anticipate before I experienced them.

"I had a nurse doing this work for nearly a decade and it was one of the best clinical models I ever ran. When I moved to locums and the nurse scribe did not come with the job, the ambient AI scribe did. The documentation got handled. The presence in the room came back. Different path, same destination."

Related resources

Blog: The End of an Era. What the Exam Room Gave Me.

Blog: The Burnout Inflection Point Is the Best Time to Incorporate

Blog: Your Patients Know Where You Live. Rural Medicine's Hidden Cost.

Free eBook: Healing the Healers: Overcoming Physician Burnout (subscriber free)

What the Research Shows About Documentation Burden

My experience is personal, but the problem it describes is systemic and well-documented at this point.

The documentation burden in numbers

• Physicians spend an average of 16 minutes per encounter on EHR documentation, with primary care physicians averaging closer to 2 hours of after-hours documentation per clinical day. (JAMA Internal Medicine, 2023)

• For every hour of direct patient care, physicians spend nearly 2 additional hours on EHR and desk work. (Annals of Internal Medicine)

• 56% of physicians cite bureaucratic tasks including documentation as the leading contributor to their burnout. (Medscape Physician Burnout Report 2025)

• Studies using ambient AI scribes have reported documentation time reductions of 50 to 76 percent per encounter, with physician-reported after-hours chart completion time dropping by a similar margin. (NEJM Catalyst, 2024; AMA Digital Medicine study, 2024)

• Physicians using ambient scribes report patient interaction quality improvements in 70 to 85 percent of cases, with patients independently rating the quality of eye contact and engagement significantly higher in scribed versus non-scribed visits. (Nuance/Microsoft DAX clinical study, 2024)

That last finding deserves more attention than it typically receives. Patients notice when their physician is present. They notice when they are not. The quality of the therapeutic relationship, which is itself a clinical instrument in ways that medicine has undervalued for a generation, is directly affected by whether the physician can give the patient their full attention or whether the documentation requirement competes for it.

How the Ambient Scribe Actually Works in My Clinic

An ambient AI scribe listens to the clinical conversation during the patient encounter, with the patient's explicit consent, and generates a structured clinical note from that conversation after the visit ends or in near-real-time during it, depending on the platform. The note typically includes a structured SOAP format or whatever clinical documentation format your EHR requires, populated from what was actually discussed in the room.

What it does not do is produce a finished note ready to sign. This is the part that every physician considering ambient scribes needs to understand before they start, because the expectation that the scribe will eliminate the documentation work entirely leads to a specific and predictable disappointment. The ambient scribe produces a draft. A good draft, often a very good draft. But a draft that requires the physician's active involvement to become a clinical note that reflects their actual clinical judgment, their specific documentation style, and the nuances of the encounter that the AI captured in language but may not have weighted correctly.

The phrase I use for this is doctor in the loop. The ambient scribe is only as good as the physician's willingness to engage with its output, correct what it got wrong, personalize what it got generically, and make the note sound like something a specific physician wrote about a specific patient rather than something a language model generated from an audio transcript. That editing process is not optional. It is the work. And it requires a learning period, a period of adjusting your own documentation vocabulary and clinical phrasing so that the scribe learns what you mean and how you express it, that most physicians underestimate when they start.

The adoption mistake most physicians make

The physicians who try an ambient scribe and abandon it within the first month almost always make the same mistake: they expect the output to be ready to sign with minimal editing, and when it is not, they conclude the technology does not work. What they have encountered is not a technology failure. It is a calibration period.

Every ambient scribe platform gets better the more you use it, and more specifically the more you correct it. The corrections you make in week two train the system to produce better output in week four. The physician who persists through the first four to six weeks of active editing and feedback almost always arrives at a workflow where the documentation time per encounter has dropped dramatically and the quality of the draft is high enough that the editing is genuinely fast rather than a second documentation effort.

The willingness to invest in the calibration period is not optional. It is what separates the physicians who adopt ambient scribes successfully from the ones who return to typing and conclude that the technology was overhyped.

The Two Outcomes That Matter Most

I had a direct comparison that most physicians do not get to make: the nurse scribe model in Plymouth and the ambient AI scribe in locums settings, same physician, similar patient volume, same documentation requirements. The two outcomes that stayed consistent across both models, and that I consider most significant, are not the ones most people emphasize when they describe this technology.

The first is restored presence in the exam room. I look at my patients during our visits now in a way that I had stopped doing without fully noticing the drift. I hear the aside at the end of the appointment that previously I might have missed because part of my attention was managing the documentation field. I can ask the follow-up question that a fully present conversation produces rather than the efficient question that a documentation-driven conversation allows. The quality of the clinical relationship, the trust, the willingness of the patient to tell me what they are actually concerned about rather than what they think they are supposed to say, has improved in ways that I believe have clinical consequences beyond the ones I can measure.

The second outcome is getting home. I realize that sounds almost trivially practical after the language I just used about clinical presence and trust. But it is not trivial. The physician who finishes clinic at 5:30 and sits at a computer for two hours completing documentation is not simply working late. They are experiencing a sustained, daily erosion of the boundary between professional and personal life that accumulates into burnout in the specific way that modern medicine produces it: not through dramatic events but through the grinding compression of time and attention that leaves nothing for the life outside the clinic. The ambient scribe gave me back those two hours. Those hours belong to Ellen, to our family, to whatever I want to do with time that is mine. The technology did not just make me more efficient. It made me a better physician and a more present person, and those two things are not as separate as the productivity framing makes them sound.

What "doctor in the loop" means in practice

The editing workflow that makes ambient scribes actually work

The ambient scribe generates the draft. The physician reviews it immediately after the encounter or at the end of the session, reading for accuracy, clinical completeness, and voice. Corrections go in at this stage: phrasing the AI got generically that you would say specifically, clinical nuances the conversation covered that the note does not reflect, assessment and plan language that needs to reflect your actual reasoning rather than a template. Over time this editing gets faster as the system learns your patterns. The physician who commits to this editing workflow for six weeks arrives at a place where documentation feels like reviewing something mostly right rather than writing something from scratch. That is the outcome. The editing is how you get there.

Related resources

• Resource: Nuance DAX (Dragon Ambient Experience): one of the leading ambient scribe platforms used in physician practices

• Resource: Suki AI and Freed: ambient scribe platform with strong primary care adoption

• Resource: Abridge: ambient documentation used in academic and independent settings

• Blog: Independent Practice Gave You Freedom. After-Hours Took It Back.

• Free eBook: Design Your Career Around Your Life: The Physician's Guide to Professional Freedom (subscriber free)

A Note for the Independent Practice Physician

If you are practicing in an independent DPC, concierge, or direct care model, the decision between a nurse scribe and an ambient AI scribe is worth making deliberately. The nurse scribe model brings a clinical presence and a verbal order execution capability the AI model does not yet match. If your volume and margin support the overhead, it deserves serious consideration. If they do not, or if you are building toward a leaner structure where headcount is a liability rather than an asset, the ambient AI scribe is the more sustainable path to the same relief. The independent practice physician chose their model partly to escape the administrative burden of institutional medicine. Whether the tool that delivers that relief is a skilled nurse or a language model running in the background of the clinical conversation, the goal is the same: the physician belongs in the conversation, not in the documentation.

For the physician building a virtual practice through a platform like ChatRx, the documentation question is somewhat different in a text-based async model. But the underlying principle is the same: the physician's attention belongs on the clinical judgment, not on the documentation mechanics. Every tool that moves the mechanics to the background and the clinical thinking to the foreground is a tool worth understanding and using.

Try it for six weeks before you decide

I am not recommending a specific ambient scribe platform in this post because the market is evolving quickly and the right platform for your practice depends on your EHR, your specialty, your patient volume, and your documentation style. What I am recommending is the category, and specifically the commitment to the calibration period that makes it work.

If you have tried an ambient scribe and abandoned it in the first few weeks, try again with a six-week commitment to the editing workflow. If you have not tried one yet, the AMA's Digital Medicine resources and the major platform free trials are the right starting point. Every ambient scribe worth evaluating offers a trial period long enough to get past the calibration curve.

The physician who comes out of the other side of that six weeks with a documentation workflow that no longer requires staying late to finish the chart, who looks their patients in the eye for the full length of the visit, and who gets home at the time they intended when they scheduled their day: that physician has not just adopted a productivity tool. They have taken back something the EHR spent twenty years slowly taking from them. That is worth six weeks of editing.

If you want to talk through how the ambient scribe fits into the broader picture of how you practice and how you structure your professional life as an owner-physician, book a $500 Business Strategy Session. The technology conversation and the practice structure conversation belong together, and this is where we have them. Join the PEA community at $99/year for Explorer membership for access to the full library of physician entrepreneur resources including the career design and burnout guides that give the ambient scribe decision its larger professional context.