trust7 min read2026-06-22

Will AI Make My Practice Feel Impersonal?

A functional medicine doctor's honest fear about AI, and a clear rule for what to automate versus what to keep human so care gets more personal, not less.

MK

Mike Kohl

Founder, Health Biz Scale

"If I automate my practice, it will feel less personal." I hear this from almost every functional medicine doctor I talk to, and I want to say up front: it is not a dumb fear. It is the right fear to have if you build the wrong system.

Your whole brand is the opposite of a walk-in clinic. Patients come to you because the last doctor gave them seven minutes and a prescription. You give them an hour, a timeline, and someone who actually reads their labs before the appointment. That is the product. If AI touches that and makes it feel like a call center, you have destroyed the thing people are paying you for.

So let me take that seriously before I argue with it.

The fear, stated plainly

You picture a patient calling with a scary symptom and getting a chatbot instead of your voice. You picture a new patient filling out a generic intake form that has nothing to do with the real story they came to tell. You picture reminder texts that read like they came from an insurance company, because they did come from that same tone of institutional nothing. You have felt what it is like to be handled instead of heard, and you refuse to do that to your own patients.

That is a values statement, not a tech objection. Good. Keep it. The rule I am about to give you protects it.

Here is what's actually impersonal right now

Before you install a single tool, look honestly at what your practice already does today, without any AI in it.

A patient calls after hours with a question. It goes to voicemail. Nobody calls back until tomorrow, maybe the day after. That is impersonal. Not because a machine touched it, because a human did not.

A new patient fills out the same six-page PDF intake every practice uses, half of it irrelevant to functional medicine, none of it read closely before the visit because you are seeing them in four minutes and you have twelve other charts open. That is impersonal.

You answer the same three questions about supplement timing and detox reactions fifteen times a week, by phone, by email, by portal message, because there is no other channel for a patient to get that answer at 9pm when they are anxious about a headache. That is impersonal, and it is also burning your best hours.

You are exhausted by Thursday because you spent Monday through Wednesday doing admin instead of medicine: chasing lab results, re-explaining protocols, fixing double-booked slots. By the time your 3pm patient sits down, you have less of yourself left to give them. That is the most impersonal thing happening in your practice, and no chatbot did it. Fatigue did it.

The system you have now already produces the exact outcome you're afraid AI will cause. It just does it slowly enough that you've stopped noticing.

What good automation actually does

Done right, AI does not replace the personal moment. It protects it.

Dr. Diane Mueller's practice is the clearest example I have watched up close. She built out systems and hired doctors under her, and the practice went from a slow trickle to roughly ten times the call volume it used to handle, with a waiting list. That did not happen because the practice got colder. It happened because the intake, the scheduling, the routine follow-up got systematized well enough that her doctors could spend their actual attention on the hard cases, the ones that need a real clinical mind, not a form.

Dan Lievens is another one. Long partnership, and one of the simplest wins was making sure patients got a response over the weekend instead of radio silence until Monday. That's not a fancier funnel. That's just: don't leave people hanging when they reached out because they were worried.

Here's the honest mechanism. You have a fixed number of hours and a fixed amount of attention. Every minute you spend re-typing a lab summary, chasing a no-show, or answering "can I take this with food" for the tenth time is a minute you are not spending on the patient in front of you. Automation does not remove the human from your practice. It removes the busywork competing with the human for your time. That is the same argument I make in When Intelligence Is Free: the scarce thing is your judgment and your attention, so protect it.

The decision rule: what to automate, what to keep human

Here is the whole rule. I'm not going to hide half of it behind a call.

Automate anything that is repetitive, factual, and does not require your clinical judgment or your relationship with the patient. Examples: appointment reminders, intake data collection, answering common protocol questions ("how do I take this supplement," "what should I expect during detox"), scheduling and rescheduling, basic lab result delivery with standard reference ranges, following up on missed forms.

Keep human anything that requires judgment, carries emotional weight, or is where the relationship actually lives. Examples: interpreting an ambiguous lab result for a specific patient's history, delivering a hard diagnosis or a scary finding, any conversation where the patient is scared or in crisis, adjusting a treatment plan that isn't working, the actual visit itself.

The test I use with clients: ask "would a smart, caring front desk person handle this the exact same way every time, using a script?" If yes, automate it. If the right answer changes based on who is asking and why, it stays with you.

One more layer to the rule. Even the automated parts should sound like your practice, not like a template. A reminder text in your voice, using your actual language about the protocol, still reads as personal even though a system sent it. A generic "Reminder: you have an appointment" reads as corporate even if a human typed it by hand. Voice matters more than who or what pressed send. This is the heart of Trust Leverage.

What this looks like in practice

  1. List every recurring question your front desk or you personally answer more than twice a week. Those are your first automation candidates.
  2. For each one, write the answer once, in your actual voice, the way you'd say it to a patient in the room.
  3. Route those into whatever system touches the patient first: text, portal, intake form, FAQ page. Let it handle the repeat questions.
  4. Watch what time that frees up. Put that time back into longer visits or same-week callbacks for patients who are struggling, the ones who used to wait because you had no room left in the week.
  5. Check every 90 days: is any automated response getting confused or generic replies from patients? That's your signal to rewrite it in your voice again, not to add more automation on top.

Where to start

Pick the single most repeated question or task in your practice this month. Not the biggest, the most repeated. Write the answer the way you'd actually say it. Put it into whatever channel your patients use first. That's the whole first step. No software purchase required to start.

If you want to build that out further, or figure out where your specific practice has the biggest gap between "feels personal" and "actually costs you patients," work with me.

Get the next essay

AI leverage, business systems, and the doctrine, one essay at a time. No pitches. Unsubscribe anytime.

Want the systems built for you? Work with me →