Transcript
YouTube English auto-captions.

========================================================================

Hey everybody, welcome to another edition of Bootstrap MD, the podcast for physician and healthcare entrepreneurs. My name is Dr. Mike Wuing. I'm a 20 plus year physician entrepreneur.

Recently, someone told me we may be the longest running podcast for physician entrepreneurs. I don't know if that's true. Maybe somebody can let me know what they think, but doing this for a long time. We have over 300 plus episodes.

You can listen to all of our episodes on bootstrapmd.com. But today, I want to talk about a scenario that might sound familiar to you. It's 9:00 at night. The kids are finally in bed.

You finished a 12-hour day. Your feet hurt. Your back hurt. Maybe even your soul hurts a little bit, too.

And what are you doing right now? You're opening up your laptop. and not to watch something on Netflix, not to finish that novel, not to decompress. You're opening up your laptop to maybe start another one, maybe even 3 hours of notes.

So, if that sounds familiar, you're not alone. So, medicine, some call it pajama time. So, this is the hours that we spent after the clinic closes, after the family's asleep, we're catching up on documentation. has quietly become one of the most crushing parts of our profession.

And here's the thing that nobody told us in medical school. We didn't sign up for this. You know, we signed up to take care of patients. We signed up to make a difference to other people's lives.

But nobody in our med school white coat ceremony said, "Hey, by the way, you're going to spend approximately 40% of your working life typing into a box. We didn't train for this. And it's okay to be frustrated about it." And here's a statistic that I recently read that physicians now spend an average of 15.6 hours per week on paperwork and administrative tasks. So, let that sink in there.

That's two full work days every single week. Two days of your life not seeing patients, not with your family, not sleeping. Document 50 weeks a year. That's a 100 work days.

roughly a third of your entire working year focused on documentation. Now, I've been to medicine long enough to watch a lot of promise solutions come and go. Ehrs were going to fix everything. Voice dictation was going to save us, but most of them just kind of shifted the burden.

They changed the tool. It didn't change the problem. So, when AI started making serious noise a few years ago, hey, I embraced it, but I was also a little trepidacious of it. We've been burned before, but you know, well, here's what changed my mind.

Like most of you, you're with an EMR and in the one that I've been using. It's rolling at an ambient AI feature. It's not a separate app. It's not a new subscription, not something I had to evaluate and procure and to convince my practice manager that we need to have this.

It was just there, just already integrated into the system that I was already using. And I tried it. And today I want to tell you kind of what happened. And I want to focus on this episode on AI scribes.

What they actually are, what's the ROI, return on investment, what it actually look like, what you generally need to watch out for, and how to get started without over complicating it. By the end of this episode, I want you to know exactly what you need to do. So, let's get into it. What is an AI scribe?

How do these tools work? Because there's a lot of confusion out there. And if you walk into the wrong mental model that nothing else will make sense. So when I say AI medical scribe, I'm not talking about that dictation software that someone must have been using for years, right?

Dragon or whatever you're using. The little microphone icon in your ENR. You speak your notes and formatting the sentences. You watch it, type it out, and it corrects your transcription.

That's transcription. That's useful, but it's not. This what we're talking about is called ambient AI. And that word ambient is doing a lot of work.

Ambient means is listening to the environment to the natural unstructured conversation of the patient encounter. It's not a formatted dictation. It's not a template. Just a human conversation happening the way real conversations actually happen which can be messy.

It can be nonlinear and ex exactly the way medicine works. So here's a picture that I want you to hold in your mind. You walk into the exam room. You greet your patient.

You have a conversation. It's natural. It's free flowing. You're not staring at a screen.

You're not typing. You're not silently composing a soap note in the back of your head while the patient's talking. You're making eye contact. You're listening.

You're fully present. You're being the doctor that you spent a decade training to become. And in the background, quietly, an AI is listening. And it's not just transcribing.

It's understanding. Knows what a chief complaint is. It knows what an assessment plan should look like. It understands clinical context of the conversation.

So, here's how it work in four steps. Step one is captured. The conversation is recorded, usually right through the app or tool already built into your EMR or through a smartphone app that's running alongside it. You start it when you're ready.

You stop it when you're done. It's unobtrusive. Patients forget it's there usually within the first minute. Step two is this transcribe and process the audience.

The audio is securely sent to the cloud, transcribed and analyzed by a large language model. The same family technology being on tools like chatbt but trained specifically on clinical language and medical documentation. It understands the difference between a level three and a level four EM visit. It knows how to structure a SOAP note.

It knows context. Step three, it generates. So within seconds, and I mean literally seconds, the A produces a fully structured clinical note, a complete subnote, progress note, whatever your workflow requires. Not a transcript dumped, an actual clinical document that you formatted and ready to review.

Step four, review and finalize. The note appears in your EMR. You review it. You edit what he's editing and you sign off.

You're in control. Remember, the AI is a scribe. You are still the physician. At least in 2020, you're still the physician.

That distinction is still the whole ball game. And we come back to it when we talk about risks. The entire goal of this technology is to eliminate the keyboard from the patient encounter to give you back your full undivided presence with the patient in the room. Think about your best moments in medicine, the ones that remind you why you do this.

And I bet most of them didn't really involve a keyboard. So what's the ROI? And I know this is how physician entrepreneurs think. A story about presence and connection, as true as it is, is not going to close a deal.

You want numbers. So let's talk numbers. The return on investment for EIC hits three completely different categories. time, money, and the human equation.

Time, this is the headline number. Multiple studies now show that the ice scribes reduce documentation by 50 to 75%. That's the physician, you know, you're talking two and a half hours saved every single day. And according to KLS research, it's a independent evaluator of healthcare technology.

The ice scribes produce a 70 to 78% reduction in that after hours pajama time I was talking about at the beginning of this podcast episode. You don't just get time back at work, you get your evenings back. That's 2 hours a day is 10 hours a week. 10 hours a week is 500 hours a year.

What could you do with 500 hours? Sleep, present your family, exercise, not physically there, but mentally still charting. Build something outside of medicine. All the above.

That's not a minor convenience. That's a quality of life transformation. Second ROI layer, revenue. Here's where it gets interesting for your practices bottom line.

And I know this is where the office manager and the CFO, this is where they pay attention. When you're rushed, when you're behind on your schedule, you know, there's throws of notes waiting for you after every patient. Documentation suffers. The complexity of the counter doesn't get fully captured.

AIC Scribe fixes that because the AI was listening to everything that were there. It's not just you having to type forever doing that. This is real money that it's getting back for you. The clinical decision making is included.

The chronic disease management, all of it, all the stuff that we usually don't have time to do and that can lead to more accurate coding, not upcoding, but accurate coding. Okay. So the financial result, what they found is a net gain of $1,300 per provider per month initially at full implementation. That's $3,200 per provider per month after the cost of subscription.

That is meaningful dollars value on there, guys. And then finally, the third ROI, the human equation. This is what doesn't show up in a balance sheet, but this is the one that matters. Physician burnout.

We all know the statistics. We all know someone. a lot of physician burnout, you know, dropped significantly from 44 to 28% after just 90 days of AA scrub use. 28% is still a lot, but it's a lot better than 44%.

But 90 days, you know, thinking about 3 months, not a wellness program, not an over organizational overhaul, 90 days of removing that documentation burden. This is important. So, where do you actually start when it comes to AI scribes on here? I want to be upfront about this and because I think it matters for how you approach this.

I'm not a product expert. I'm not here to stand up and give you a rank list of every AI scribe in the market because I haven't particularly used all of these and many of you have and could probably speak a lot more on this than I can. But here's what I tell colleagues. Start with what you actually already have before you research a new single product or sign up for a single free trial.

Open up your EMR and it may already be there. You know, that's what I did. My EMR already had an ambient AI feature built in. It already integrated.

It was already part of my existing workflow. I didn't have to get a new subscription, a new app. You know, that can take time on their time away from it. It's already built in.

Athena Health is popular. They've been rolling out ambient AI as a native feature. Epic the ambient AI functionality. Other major systems are already started or already following suit soon.

The EHR offenders have finally figured out this is a feature that users actually want. So the first two questions ask, do you already have ambient AI or AI scribe feature available and how do I turn it on? You might already be paying for something that you're not already using. So the first place is to look.

Now if your EMR doesn't have it yet or maybe what they're offering isn't good for your workflow, there are a lot of standalone tools worth exploring. Patient notes, a free.ai, a bridge, deepcribe, suki. These are names that I've seen come up frequently and most of them honor offer free trials. Now I will be honest with you.

I haven't personally used all these tools at depth but I know the reputation and you know ask someone who's used these different tools who is in your industry and they could probably point you in the right direction. But I want to focus on the one thing that matters when you're evaluating any tool on in there is how does it fit into how I'm actually seeing patients. Not how the demo works, not how the sales rep describe it, but how it works when you're in a room with a complex patient and you're 15 minutes behind and the note needs to be accurate. That's the test and the only way to answer it is to actually try it.

Which is why I want to walk through this with you here. You know, I'm really enthusiastic about this technology, but I would be doing you a disservice if I glossed over what can go wrong. And there's three real risks, but it's all manageable, but you can't ignore it. Risk number one is accuracy and hallucinations.

Gosh, I've seen this, but it's getting better. AI scribe notes are often better than what a rushed, exhaustive physician writes at 9:00 at night. That's the honest truth, but they're not perfect. The technical tool for when an AI confidently generates inaccurate information is called a hallucination.

In clinical note, that can mean a detail that was never said, a medication that wasn't mentioned, a symptom the patient didn't report. Those are now omission errors where the AI simply misses something are more common than fabrication errors. But both can happen and in medicine not impossible is not an acceptable standard. So here's the golden rule and say it with me.

The physician at least in 2026 is always the final editor and the legal signatory of that note. The idea has just described you are the physician but if something goes wrong it ends up in the note and you signed it. That is in your chart. It's your liability.

It's your license. the AI company is probably not going to show up at your deposition on here. You know, I don't want to just gloss this over, but again, it is a tool. You are the physician on here.

So, treat every AI generate note the way you would treat like a third-year medical student. You know, you're grateful for the draft, but ultimately the attending or physician is responsible for what you sign. That review step is not for meldy. It is practicing medicine.

Risk two, patient privacy in HIPPA. You're recording a patient encounter that carries real, legal, and ethical weight. On the consent form, you must inform your patient that the visit is being documented with an AI assistance and get their consent before you start. It doesn't have to be complicated.

Sign at the door. Send this at the start of the visit to give you my full intention. I'm using an AI assistant to help with my documentation. Is that okay with you?

Then document it. One line in that chart. Patient informed of AI documentation assistant and provided verbal consent. Done.

Now on vendor, if you're using a standal loan tool, not your EMR's built-in feature, the vendor must sign what's a BAA, which is a business associate agreement. That's that HIPPA contract that makes them legally responsible for how they handle protected health information. If a vendor won't sign a BAA, it's time to walk away. Now, if the EMR is the AI is already built into your EMR, the HIPPA is already handled through your existing relationship with your EHR vendor.

So that's definitely an advantage that you need to be aware of on here. Now risk three is the billing dilemma. More complete documentation often leads to more accurate and sometimes higher coding. This is legitimate when it reflects genuine medical complexity.

But there's a line you don't want to cross. Uping. What's up coding? You know what it is.

Building for a higher level of service than what was medically necessary and documented. That is fraud. Regulators are watching this space. If your billing pattern shifts dramatically overnight after implementing AI, that could trigger a payer audit.

So use the AI coding suggestions as a starting point, but it's not a final answer. You need to apply your clinical judgment. Does the documentation actually support the code level? Does the medical necessity justified?

If you spill it, if it doesn't, don't do it. More complete notes should produce more accurate coding, not automatically, more expensive coding. The difference is your judgment. And your judgment is what your license is built upon.

All right. Finally, I give you a five-step implementation plan. We've already covered the what, the why, the where to start, and the risks. So, I'm going to give you a five-step plan so you can actually get this off the ground.

All right. Number one, ask your EMR first. Before you research anything else, contact your AMR support. Do you have ambient AI or AI scribe feature?

Is it coming? What's going on? If you already have it, how do I access it access it? You might already have access to something you're not using.

That's that's the first goal. Everything else comes after. Step two, run a 30-day pilot. Don't roll out anything to your entire practice on day one.

Maybe if you got more than one physician, have them do a 30-day trial long enough to get through the learning curve. Sure enough, that doesn't work. The cost was minimal. The pilot will surface things you cannot predict in advanced edge cases.

The AI handles poorly in your specialty. The workflow friction points. The patient question your front desk isn't ready for. Get those lessons from two people before you scale.

Step three, get patient consent right before you go live. Before a single recording happens, build a consent workflow. A sign on an exam door, a scripted sentence the physician says at the start of every visit, one line of chart documentation confirming consent. Make it as automatic as washing your hands before you walk into a room.

Again, reframe how you think about it. You're not disclosing a liability. You're telling a patient that you care about being present for them. Most patients respond positely.

I've had a few who said no. That's all right. you know, but it's a lot better than what having a keyboard in front of taken for someone who was a patient and the doctor was using your keyboard. It's very impersonal.

This is much better. Step four, budget time for review in month one. This is the step most practices shortcut and where most problems start. So, for the first month, budget an extra 5 to 10 minutes per note to actually read the AI output, not skim, just read it every note, every field, verify what's there, correct what's wrong and what's missing.

in month one that eats into the time saving. That's intentional. You're catching errors before they become chart errors and every correction you make teaches the AI your documentation style and the investment in with one is going to pay off for years to come. All right.

Step five, measure before and after. Before your pilot starts, track your baseline documentation time per day after hour charting hours per week, patient throughput. Even a simple 1 to 10 burnout rating, anything is better than nothing. and then reme-measure it at 30, 60 and 90 days.

Remember, if you want to change something, you need to measure it. Okay? So, data is your currency collected from the beginning. The AI scribe revolution is not coming.

It's already here. And physicians who are using it, they're reclaiming two to three hours a day. That's adding meaningful revenue per provider. They're watching their burnout numbers drop.

They're being fully present again with their patient, which is why most of us went into medicine in the first place on here. It's not perfect. And again, what we're talking in 2026 could be a lot different 5 years later from now. All right?

So, let me know how AICRE works for you. Try it out if you haven't done it before. It's going to save you a lot of time, money, and energy. Guys, you're going to have ups and downs as a physician entrepreneur.

Do something a little each day to get you closer to your goals and keep moving forward.