Beyond the hype, AI is already impacting how practices attract and convert patients. Jeff and Dan walk through actual applications you can use immediately to improve marketing, efficiency, and new patient flow.
DDS Success (coupon code RX269) - https://ddssuccess.com/
Blumberg Digital - https://www.blumbergdigital.com/
Listen to full episode :
Have a question for Jeff?
Fill out the form and he will get back to you.
Questions From This Episode
What is an AI agent, and how are dental practices using one?
An AI agent is a preset series of instructions that runs step by step, choosing the next step based on the results of the last, and does it perfectly every time. The clearest example in dentistry is the AI receptionist, which answers after-hours and overflow calls, screens for emergencies, logs the details, and alerts the office manager by phone, text, and email.
Why add an AI after-hours receptionist?
Dan cites a study that 80% of callers who reach a practice voicemail hang up. An AI receptionist answers those calls instead, and Dan suggests it can pick up about two new patients a month. Because ads cost about a third to a fourth as much outside office hours, practices can also advertise in the evening and on weekends. Dan stresses that it should supplement your team, not replace it.
How can AI help with reviews and social media?
Paste a practice's patient reviews into an AI tool and ask it to extract what patients mention most, which Dan calls "buttons." For one client, 45% of reviews praised friendly, professional staff. Dan did the analysis of over 900 reviews in 15 minutes, then used the results to draft a month of social media content in under an hour.
How can you find out what AI thinks of your practice?
Ask ChatGPT, Gemini, or Grok which dental practices in your area they recommend, why, and what your practice is missing. Dan notes that Google is still the backbone of the local data these tools use, so you should not drop traditional marketing. Use the results as a shortlist to share with your marketing company.
What should dental practices be careful of with AI?
Be wary of any claim that AI can handle all of your marketing, because Dan's trials produced content that violated state dental practice acts and promoted services the practice does not offer. Also challenge HIPAA compliance, both for the platform and for the AI provider behind it, and avoid giving AI agents access to sensitive data like banking and email.
Episode Transcript
-
Jeff: News about AI and how it's going to change the way you, me, and everyone else does business is everywhere, from financial markets to diagnostic tools to marketing. Well, just about everything. But beyond the excitement, worries, and hype, what's actually happening? What AI tools, especially in relation to marketing or practice management, are actually useful now or soon to become useful, and which appear to be more hype than anything else? And how is AI changing or affecting marketing and new patient acquisition?
For this week's episode, we're going to do a deep dive on these questions and more. Rather than a general informational update, my plan is to focus on things you can leverage or do right now with these tools to improve outcomes in your practice.
My name is Jeff Blumberg and I'm your host, and I'm joined this week by Dan Brown. If you've been listening to the podcast for a while, you know Dan. Dan is the COO of Blumberg Digital and a senior seminar speaker here at MGE. Blumberg Digital is the premier digital marketing agency in dentistry. They know how to get results, and while yes, it shares my last name, I am not a principal of the company, although I highly recommend them. I'm super happy to have Dan on. He's a leading authority on digital marketing in the dental space, and every time we talk I learn something new. Like me, Dan was an early adopter of AI solutions and saw the writing on the wall as to the potential effects, good and bad, it would have on dental marketing and the administration of a dental practice. He has really specialized in these things, while I, on the other hand, have been more focused on how to use AI to troll my brother-in-law. Just kidding, but not really. My focus with AI has been more on analytics and using it to streamline workflows. If you're a dentist looking to fix or improve your marketing, Dan's experience and knowledge are super relevant, and I'm truly excited about this episode and the potential it has to help your practice. With that, let's bring Dan on. Dan, welcome to the show.
Dan: Thank you. Nice to be back.
Jeff: AI. It's crazy how people are losing their minds over this. And I find some people are still saying, "I don't know if I really want to use AI. I don't like the idea. It scares me."
Dan: It's interesting.
Jeff: Well, I have to say I love AI, because when they eventually take over...
Dan: We both love AI. Our family loves AI.
Jeff: My bank loves AI. I know it's been affecting financial markets pretty heavily the last few weeks, especially software companies. People are afraid, because everybody's buying cloud-based software, and financial markets are concerned that AI will let you customize software in a matter of minutes so you won't need these products anymore.
Dan: And it's actually happening.
Jeff: Okay, so see, this is why we have you on. Why don't we just jump into it. I know we're going to stay away from the diagnostic stuff, like tools that read X-rays. I'll leave that to clinicians to determine whether it helps them or not. I've heard promising things about it. But here's the first question. Could you give me a brief rundown of what's out there right now for practice management or marketing help in the dental space, generally, without getting into brands, and your opinion on where each of these is at?
Dan: Sure. One of the first things anybody should start doing right now, if they're not already, is jumping in and playing around with AI. That means getting an account with OpenAI for ChatGPT, or with Gemini, and starting to build a profile. There are little courses you can take to learn what AI is, what it can do, and what some good use cases are. I do suggest doing a basic course on it. There are a bunch out there. If you start searching for a course to learn AI, your feed is going to be absolutely slammed with them. Find something that costs a few bucks and just learn.
Jeff: I think the subject has to become less scary as soon as possible, because it's going to be so dominant. It already is in a lot of fields, but in the coming years you don't want to be behind that wave. It reminds me a bit of the mid to late '90s. Some people will say they weren't born then, but I was. The first thing I started using was CompuServe and then America Online. My original email address was something ridiculous you can't get now. And then you had the craziness toward the late '90s with all the different internet companies. A lot of people said the internet was going to do this or that, and none of that happened. But there were a lot of things nobody predicted it would do.
Dan: There's definitely that, but I would say what's happened in the last two months makes this fantastically different from that period. We don't need to go into the weeds, but OpenClaw was released. OpenClaw is what's called agentic AI, or an AI agent. Instead of giving an AI a prompt and then the next prompt, you have a series of prompts, and you can integrate with your software, your Gmail, whatever platform you use in your workplace.
Jeff: I don't want to get too technical, but let's cover the basics. Most people have heard of or have used ChatGPT. The thing you type when you ask "What's a good movie?" is a prompt. Obviously it can become more complex: "Design this. You're this." You're talking about AI agents. Can you give a brief primer on those?
Dan: Yes, and even if you have no intention of moving into AI, it's important to know what this is. An AI agent is a preset series of instructions to do step one, and then based on the results of step one, you go to A, B, C, D, or E. Once it goes to B, it has another series of instructions, and based on those results you go to the next series of choices. All of this happens in seconds. It lets you do a series of things in sequence according to your usual, normal procedure, and it does it repetitively and perfectly every single time, provided you give it those parameters.
Jeff: Can you give an example of using one?
Dan: A perfect example is the AI receptionist. It's a service we put in place with Blumberg Digital. A proven study shows that 80% of calls that hit a voicemail in a practice are hung up on. Say somebody calls a practice after hours, on the weekend, or during lunch, because sometimes practices close for lunch. When they hit a voicemail, 80% of those callers hang up. Anybody in a practice has probably seen the pitch "You need an AI receptionist." We created one based on the principles of MGE, DDS Success, and Blumberg Digital, with exact instructions for the circumstances under which calls come in and what to do in each. If it's an existing patient, it does this. If it's an emergency, it does this. If it's a real medical emergency, it does this.
Jeff: It tells them to call 911, I assume.
Dan: Call 911, and then it alerts the relevant people in the practice. It sends an email. It can route the call to the doctor's phone. Based on a series of variables, it does something different each time. It's not just answering the call and saying something. I have a sample we could even call right now.
Jeff: A sample phone call? Sure, let's do it. I'll be any old patient. I wonder if they'll be able to hear. Let's try it.
(Sample call begins. Dan plays the caller.)
Lindsay (AI receptionist): Hello, you've reached Bright Smiles Dental in Clearwater, Florida. I'm Lindsay, the AI receptionist. How can I help you?
Dan (caller): Yeah, I just broke a tooth and I don't know what to do. I know it's after hours, but I'm in so much pain and I just need to see if I can get in at the next possible time.
Lindsay: Oh no, I'm really sorry you're in so much pain. Let's get this marked as urgent for our team. First, can I get your first name please?
Dan (caller): It's Daniel.
Lindsay: Thank you, Daniel. And your last name?
Dan (caller): Brown, like the color.
Lindsay: Got it, Daniel Brown. What's the best phone number for our team to reach you? If you'd like me to use the number you're calling from, just let me know.
Dan (caller): The number I'm calling from is perfect.
Lindsay: No problem. I'll use the number you're calling from. On a scale of one to 10, how severe is your pain right now?
Dan (caller): I'm not really good with pain, so I'm sitting at about a solid eight.
Lindsay: That sounds really tough, Daniel. How long have you been dealing with this pain?
Dan (caller): Since last night, actually. I bit down on something hard, and I heard a crunch. Something snapped. The tooth had been a little sore for about two months before that, but something went wrong last night.
Lindsay: Oof, that sounds rough. Which tooth or area is hurting the most right now?
Dan (caller): On the right-hand side, on the top, toward the back.
Lindsay: Thanks for sharing that, Daniel. Has anything helped or made the pain worse since it started?
Dan (caller): I had a cold drink, and that sent me into the stratosphere. I almost passed out. I can't have anything cold in my mouth right now.
Lindsay: That sounds absolutely miserable. I'm really sorry you're going through this. I'll mark your case as urgent and let our team know you need the next available appointment. What day would you prefer to come in, or do you just want the soonest possible?
Dan (caller): As soon as possible. I'll make tomorrow morning available. It's important to me. I'm in town for my sister's wedding on Sunday, so I'm really hoping to get the soonest.
Lindsay: Absolutely, Daniel. I'll note that you're hoping for the soonest possible, tomorrow morning if available, since you have your sister's wedding coming up. Our scheduling team will see this as urgent and reach out as soon as possible to confirm your appointment. Is there anything else I can do for you right now?
Dan (caller): Well, on the subject of my sister's wedding, I've heard there are things you can do to make your teeth whiter. I'm going to be in the wedding party, so I don't know if you have such stuff.
Lindsay: Oh, totally get wanting to look your best for the big day. Yes, we do offer teeth whitening here at Bright Smiles Dental. Our team can talk with you about options when they call you back, especially since you're already coming in for that urgent tooth issue. I'll make a note so they know it's important to you. Anything else on your mind before I pass this along?
Dan (caller): No, that's good. I appreciate it.
Lindsay: You're welcome, Daniel. I hope you feel better soon. Our team will call you back as soon as possible to get you in. Thanks for calling Bright Smiles Dental. Hang in there, and good luck at your sister's wedding.
(Sample call ends.)
Dan: So I've ended that phone call.
Jeff: Now, if I could just comment. There are two things I really like about this. One, it's a lot further along than I thought this stuff was, as far as recognizing what your issues were. And two, I like the way you're deploying it. I've seen other companies deploy AI receptionists with the idea of fully replacing your receptionist. You're deploying it for when you just can't get to a call, which is not ideal, but after hours and on weekends, that stuff drives me nuts. I'll look at a client's Google pay-per-click dashboard and see they got 15 calls over the weekend, or they close at 4:00 and there were 10 calls between 4:00 and 8:00 PM.
Dan: One hundred percent. And if you can see what's happening on my phone right now, it's calling me, because I'm the designated contact as the office manager. This is a sample, and here it goes.
AI receptionist (call to the office manager): Hello, Dan Brown. I am Lindsay, your AI receptionist. We have an emergency patient, Daniel Brown. They have a broken tooth on the right upper back. Severe pain, 8 out of 10, worsened by cold. Urgent care needed. Please call them back at...
Dan: I'll leave the number out. And then this text just came through on my phone, as the office manager. Patient name, Daniel Brown. Phone number, which I'm not going to say. Reason: broken tooth on right upper back. Severe pain. Urgent care needed. Please call back ASAP. And I also just got an email with all the data about everything.
Jeff: Wow, look at that. Service requested, how they heard about us, appointment time, emergency, call date. That's intense.
Dan: We call this overflow, if the team's just busy. Or lunchtime coverage, or meal coverage.
Jeff: And the brilliance of it goes back to what I said about Google pay-per-click. I see a bunch of calls missed. A lot of listeners know Rushi Trivedi. He has two practices up in New York. One of the things he did years ago, before AI was possible, was look at how many calls he was missing after hours, and he hired people to answer them. People are probably missing new patients because they don't have something like this. And if you've got an ads company, they're probably turning off your ads during your non-working hours.
Dan: Right. Yet during your non-working hours, the ad costs are about a fourth.
Jeff: So how much am I paying? Say I'm advertising in a pretty competitive market, during my office hours. How much am I paying per click?
Dan: For a common term like "dentist near me," it's going to be $20 to $30 depending on the area.
Jeff: What about implants?
Dan: Implants are $60 to $90 per click during your office hours. That goes down to about $15 outside of them.
Jeff: Wow. So now we're advertising more heavily in the evening and weekend times, and we're using the AI receptionist.
Dan: An AI receptionist also goes really wrong when you don't say it's an AI receptionist. People get upset, because they notice a slight delay or a slight mispronunciation. It's not perfect. But we use the top of the top in text-to-voice. There are different companies like ElevenLabs and Synthflow, and we use the best of the best. Then we create what's called a logic center, or knowledge center, and program in all the parameters from DDS Success and MGE. So it's acting like the most perfectly trained receptionist. She's friendly and appropriate, and then it does the alerting and the logging. We can tie it in with their software.
Jeff: That's better than an answering service. With an answering service it's "Hello, Dr. So-and-So's office. The office is closed. Do you need us to reach out to the doctor?" There's no data. Are they a new patient? Are they a patient of record? Most answering services don't do this.
Dan: In the last four years we've worked in tandem with a few different answering services for after-hours stuff and coverage, because you can lose your receptionist for two weeks unexpectedly. We try to find one that's consistent, appropriate, and quick, and it's never less than $4,000 a month. Even the one we finally said we could refer to our clients without it coming back on us, at $4,000 a month, something happened in the company, things went down, and the live operators just weren't good. This AI handles 33 languages, so if you start speaking another language, it just flips to that language. And it's dirt cheap. If you own an answering service, it's not a good business to be in right now.
Jeff: Probably not. But there will still be people who insist on a live person. That's just my opinion. If I had an employee who had access to the software and could schedule patients, sure. But this was much better than an answering service. It also winnows out whether a call is worth acting on. What if somebody calls up and says, "I'm in really bad pain," and you ask, "Are you a patient?" and they say, "No, I just need some drugs"? That happens. They don't want to come in, they just want a prescription over the phone, which you're not going to do.
Dan: That's right. And we actually have one client that lost a receptionist unexpectedly, and for a week and a half the AI covered it brilliantly. It routed the calls and screened things. Our next stage is outbound. We'll have it send texts, emails, and phone calls to get people on the phone and transfer them to the dental practice, because who has time to put three staff on the phone to call 4,000 patients?
Jeff: That's the big problem, especially when you look at reactivation. I analyzed a practice and went over it on the podcast a few months ago. I forget the exact numbers, but roughly it had 5,000 charts, and 700 people had a next visit date. This was within the last two or three years, not from 1987. If I put somebody on the phone, how many calls are they going to make a day? If they're really good, a superstar, 100 or 120. The average person is going to do maybe 50 to 70. Contacting 5,000 people would take months. But if you could deploy this to call 5,000 people at once, or 4,300 in this example, that's a whole different beast.
Dan: We're going to scale it so that it says, "I'm the AI representative for ABC Family Dentistry. I've been trying to reach you. I've got Megan right here, standing by. Can I transfer the call?" Then it transfers the call to somebody in the office. This would be done during normal hours, and we'll have different programs for different things. Sometimes an office just doesn't have the resources, so we'll take it as far as we can, doing reactivation and messages, and then use-it-or-lose-it benefits toward the end of the year.
Jeff: This is cool. We'll put your email address and a link to your site on the episode webpage, so if people have questions they can ask you about this.
Dan: Perfect. This is being used right now.
Jeff: And again, I like the methodology you're using. It's not "let's replace a human." I know I'm saying this a lot, but this is where I see the selling point, because people already have staff problems. It should be used when you don't have a human.
Dan: Right. We're not yet at the point where it's seamless and smooth in every case. It's taking a call and routing it to the office manager.
Jeff: Great. So that's one.
Dan: You asked for real-world things you can do. We're using AI to take reviews from Google, Yelp, BBB where applicable, Nextdoor where applicable, and so on. A big push of our marketing company is to get a volume of reviews. Some of our practices have 300, 600, 900, 1,200, 1,600, even 2,300 reviews. We go in and copy the reviews, expanding them where it says "more," paste them into AI, and extract what we call buttons. Basically we run a survey: what do people like about this practice?
I have an example from one of our clients that has a very successful number of reviews. I did an analysis asking what the reviews say and what the buttons are. That comes from the consumer, and we're always trying to look at what a consumer thinks about dentistry and about our practice. You have to think like a consumer in order to communicate to the consumer. So we extract the things people mention most. For this particular practice, 45% of patients say friendly and professional staff, 38% say clean and modern facility, and 32% say painless procedures. A practice can't say "painless," but we can now extract those reviews, caption them on social media, and build a series of promotions and marketing campaigns around patient testimonials. Consumers don't want to hear what the practice thinks about the practice. They want to hear what patients think about it. This particular practice has over 900 reviews, and I did this analysis today in 15 minutes.
Jeff: That would have taken hours and hours. I think that's an immediate use case everybody can apply: analyzing data. For somebody to read 900 reviews and collate them, you'd have human error. Did they say "friendly and professional staff"? Did they say "happy staff"? It would get miswritten. I've seen those survey tabulations, and they're a nightmare.
Dan: And now I can turn around and tell the AI, "I'd like to build campaigns around each of these buttons. I'd like social media text. This is my area, and this is the service I want to promote." It will pump out 20 social media posts. It will even create imagery if I feed it our logo. A lot of people don't have time to do social media posts, or they wonder how to make them communicate. Now you've just heard what's most important to your patient base, and you want more patients like those. In about 45 minutes to an hour I can have all of the month's social media content, which would have taken days in the past.
Jeff: Who has time to sit down and click around in some online Canva-type tool? I don't like doing that. I'm not a creative by nature, but I can tell something that looks great from something that looks janky, and this will pump out some pretty nice stuff. So we've got data compilation and the AI receptionist. Any other things being used or promoted right now, again not diagnostic, for practice management or marketing?
Dan: I use AI to do competitive analysis of neighboring practices. Everybody should do this. Log into ChatGPT and Gemini and ask, "What are some of the great dental practices in this area?"
Jeff: Would they use their ZIP code or their city?
Dan: Either. Some cities are too big, so you want to narrow it down to what's colloquially known as your area. You want to find out what AI thinks about the top practices and note those down. Then ask AI, "Why are you recommending these?" And at the end you can ask, "Why aren't you recommending my practice? What am I missing? What do these practices have that I don't? If I put that in place for my practice, ABC Family Dentistry, how could we be regarded more highly by you? Please walk me through that." Then watch the instructions come through.
Jeff: Interesting. When all of this started a few years ago, with the deployment of ChatGPT, Grok, Gemini, I remember thinking I'd still use Google for my basic search, and I still do, because AI didn't seem as plugged in to exact local search. Google is still kind of the king there. But my wife, my daughter, other people I know are using OpenAI and ChatGPT: "Hey, what are the good restaurants in this area?" instead of going to Google or TripAdvisor or Michelin. It's so much more conversational than Google, where you get your 10 results and still have to go through them. Now you can have AI go through them. Now, between you and me, your wife is highly intelligent, you're highly intelligent, your daughter is highly intelligent, and progressive. There's still a vast number of people who aren't like that. So this is what I want to say to everybody: don't run out and stop all of your traditional marketing and think you're missing tons of opportunity because you haven't done these steps.
Dan: Unfortunately, only about 4% of the market is that progressive in its thinking.
Jeff: Also, ChatGPT and even Gemini, which is owned by Google, where are they getting their data? From Google.
Dan: Right, and ChatGPT will openly say that, and Grok will openly say that. Google has amassed the largest database of local business information and all the reviews. It's still the review king. Google is the backbone. We're just using another interface to talk about what's already on Google. If you're a premium SEO client with Blumberg Digital, we have what's called an SEO bolt-on for AI. It's the supercharger on top of traditional SEO, where we do exactly this. We find out what AI thinks about your practice, and then we kind of placate AI, because the AI companies aren't yet lifting the hood and telling us what to do to rank well. So we have to reverse engineer it until they give us a best practices sheet.
Jeff: As you're doing this, are you finding a positive path forward? Like, if we do this, it seems to rank better with an AI platform?
Dan: One hundred percent, until they change it. Google isn't going to sit down while ChatGPT takes all this market share, so I think we're all in store for some big surprises from Google. ChatGPT was the industry leader, then Gemini came in hard, then ChatGPT made big changes and dropped certain programs. That works for us as consumers, because they're all working to get better so we want to buy their stuff.
Jeff: So say I'm a general dentist and I'm not really using anything. Maybe I use ChatGPT to help me write policies. What AI tools are ready to be used right now? I can do this AI reception thing. I can potentially use it for reactivation in the future, so I don't have to hire five people to call my whole patient base. I can use it to determine what I need to do to make my practice more appealing, and I can use it to compile marketing data, or have you guys do that. Is there anything else you would have me leverage that's really usable now?
Dan: I like NotebookLM, and we're building something like this right now. I don't know if you've heard of it.
Jeff: No.
Dan: NotebookLM is a Google tool where you can drag in documents, conversations, and YouTube videos. You don't even have to drag them, you just put in the link. You create a sub-mind, so to speak. I have it open here. It takes a series of documents, recordings, or policies, groups them as a database, and then you query that sole source material.
Jeff: So it's like a personally searchable database. I could take all of the MGE blog and all of the MGE YouTube videos, create a searchable database, and ask it a question.
Dan: That's right. It's still nascent on my end, but we're doing that with everything from me and Sabri, so there's an interactive AI agent where someone can say, "Here's my practice, here's what's going on," and it spits out, "Here are your top problems, one, two, three. This is what you should be working on right now."
Jeff: So the key is the validity of the data.
Dan: Yes. One big problem with AI is hallucination. For the listeners, that's when AI grabs and invents data to placate the user.
Jeff: Like that lawyer who cited all those cases that didn't exist using ChatGPT.
Dan: One hundred percent. When I was using AI to help co-create our AI receptionist page, it spit out 15 testimonials.
Jeff: That's not scary at all.
Dan: I don't know if you know this, but Meta, with Facebook, now has an AI platform they haven't deployed yet that can take someone who has passed away and continue to post, answer, react, and share posts on that person's Facebook page.
Jeff: Scary. That's messed up.
Dan: So here's what we did with one of our clients. We found some of their most successful implant lead calls, the first calls that came in. Then we took their initial virtual consultations, which we do for implant leads on a HIPAA-compliant platform. We took 10 successful first phone calls and 10 successful closes, meaning the initial virtual consult resulted in the patient coming in and closing.
Jeff: So you had the transcripts of all of these?
Dan: We just used the audio. We dumped it all in and created reverse-engineered training manuals, scripts, and prompts. So when it comes to "what about this or that," it's only grabbing that data. It's not hallucinating, inventing data, or grabbing data from a non-authorized source, which is what AI does. If I'm running a practice, I want it to be MGE brand, DDS Success brand. I don't want it to do what other practices have done.
Jeff: So if I understand correctly, you take the audio from 10 implant calls where the receptionist got them into a consult, and 10 successful consults, you drop it into NotebookLM, and now I can ask it questions. One question I should ask: does Google have access to my stuff to reproduce and use it, or is it firewalled off for myself?
Dan: It says it's firewalled off. But I'm here to tell you I trust nobody and nothing.
Jeff: We agree on that point. I wouldn't trust anything.
Dan: I want to circle back to this big surge of people running out and signing up for OpenClaw, this agent creator. I have it set up on a server on Amazon with no access. People are plugging in their iMessage, their Gmail, their bank data, their Chrome data, full access to everything. And what's great is, "When I wake up in the morning, please have all my emails answered with proposed replies based on my history. Please give me a summary of all my bank balances and my meetings." It's beautiful. It's amazing. But you've just given something access to everything, which means tomorrow morning you could have nothing in your accounts. All of your emails could have been answered by some ghost, and it's destroyed your life.
Jeff: I probably shouldn't be laughing about that. Well, we laugh out of "Oh my gosh, is that true?" I wouldn't add any sensitive data. In this case, I'm dropping in basically transcripts, and that's all I'm doing.
Dan: Right. Now it produces workflows and policies for you: this is how you answer the phones, this is how you do this. Then, if I'm a trainee taking these calls, I can ask, "What do you do when a patient says this?" and it gives me an answer based on that training. It will also produce quizzes, flashcards, and diagrams.
Jeff: I've used various platforms for training like that. I'll have a policy and say, "Give me five drills based on this policy."
Dan: The key is in the prompt. It's what you ask it to do. If what you ask is weak or has holes in it, you're not going to get a good response. I've been working on prompt engineering since this all started, about three years. For the person who doesn't know how to prompt, NotebookLM is a nice solution. It's an AI agent. It will create a quiz, flashcards, and diagrams, and some of them might be okay and some might not be. That's fine. I edit it. "I don't like that quiz question, let's change it to this. I think it misunderstood this, let's fix that." It gives you a first step. It rids me of 90% of my work.
Jeff: I've done videos and episodes on this. When I mentioned AI in the first AI episode I did, I got the most engagement I ever had, but it was trolly as all get out. "AI's going to take all the jobs." No, that wasn't what I was talking about.
Dan: Even this reception tool isn't meant to take jobs. It's answering calls 24/7. Imagine having that amazing, high-end team member who is currently a receptionist now doing follow-ups on consults and making calls for you, doctor. This lets us take resources we have that are otherwise overloaded. What's the alternative right now? We get slammed at lunchtime, we put the phones on voicemail, and I lose five new patient calls. Now at least they get answered. Of the 20 practices we piloted this on, all 20 have retained the service and said they would never live without it.
Jeff: I believe it, because the alternative is an answering service that's just going to say they're closed, maybe not rudely, but on average that's what's going to happen. Or I might get a voicemail, in which case I just blew that opportunity with a new patient call. Unless it's a patient who really is already a patient in my practice, they might leave a voicemail.
Back to the episode I did: it was how to use AI to write policy for your practice. Most people say, "I don't have time to do manuals." I gave them a little instruction with some prompts. Record what your financial policy is. For example, your patient can prepay and save 5%, or put half down and half later, whatever that might be. Then take that transcript, and I gave them the prompt to use to write the policy, and it creates a policy you can give to your team. That was the initial leverage point for people listening. And of course people thought I was trying to get AI to take the jobs. No, I was trying to improve the people who were there. None of the listeners commented on it. It was the general public, trolls.
Dan: Now a person can, while driving to work or while training a new staff member, just hit record, drag the recording off their phone into Google Drive, and be done.
Jeff: That's what I've done numerous times. I use something like Plaud, get my transcript, drop it in, and tell the AI what I want it to do. "Turn this into a policy. Give me three exercises based on this." The trick used to be that most software's transcription was terrible. That seems to be fixed across the board. I've used it to write manuals and all kinds of stuff, and that was the first use case I promoted. If you don't have a manual, you now have no excuse.
Dan: A person is training verbally, and then the person they trained gets promoted, and now you have to do it again for the new incumbent. Or someone quits, or has a family emergency and has to leave suddenly for six months. The next time you verbally train another person, just start recording. You can record video or audio.
Jeff: I agree. Those are the main things right now for practice management and marketing. Now, what do you see coming in the next year, two years, five years? This is spitballing. Not globally, where robots are going to be everywhere, but how do you see this affecting dental practices?
Dan: I think agentic AI is the big deal. The guy who created OpenClaw, the one I told you about, was just headhunted by all of the top three AI companies, and he signed a billion-dollar sign-on bonus. I don't know if I said million. It was a billion.
Jeff: It's even better than LeBron's.
Dan: It's unheard of. It made the news. What does that mean? He created this thing, it cost him $20,000 a month for months, he released it as open source, and it got him a job. That tells you where AI agents are going and how many people are using them. Which is a scary thought, because I give it access to everything. But now imagine this. I tell the agent, "These are the recordings I've done. I want you to create a series of manuals, quizzes, and training paths for all new staff in each of these positions, organized this way. Set up a training program for all of my new staff. Create a logging system where they can report any questions they have, and send me those as a daily report so I can see a summary of all my staff. Check in with my staff: what was great about today, what could have been better? Propose new policy." That's agentic AI.
Jeff: So OpenClaw basically made it easy to create an agent, a series of agents around a focus or an environment that handles everything as your personal assistant. I've created agents on OpenAI, which takes a little more thinking and doing. This sounds like an easy interface.
Dan: Exactly. That's where things are going. And Elon Musk just closed down production of the Tesla Model X and Model S, and he's using those factories to create robots, which all use AI agents. I know this sounds like Blade Runner, I, Robot, Terminator, but it is going there. It could be a robot that is with your elderly parent or grandparent and cares for them, because caregivers are known to be abusive, neglectful, and expensive. This will be at a fraction of the cost, 100 times better, and it can immediately alert the hospital, the doctor, or the kids or grandkids. This is where it's going. Hopefully not to replace us or attack us, but to facilitate us so we can put our attention and talent where they can be used best.
Jeff: To riff on what you're saying, I'd guess robotics would first step in where things don't require fine motor skills. If I load ships for a living, you could probably get robots to do that, because you're not going to hurt a human, you're just going to break a package. They might sweep streets and pick up trash. There are even robotic oil drilling rigs now. But caring for an elderly person, childcare, or being around people is a whole other thing, because all it takes is for one of those things to screw up and hurt somebody, which you know is going to happen.
Dan: So what should we focus on today? The menial things, the things you have to do over and over again. If they don't get done, it really causes problems, but they don't take a lot of skill, and variation in them is frustrating. If you could eliminate the variation and the repetitive nature of it, what would that free up? Answer that for yourself. What in your life doesn't require all of your intelligence and attention, but if it doesn't get done, is really irritating? Start making a list of those things now, take the easiest item on the list, then the next, then the next. Take some of those functions off your plate, or your office manager's, your new patient coordinator's, or your insurance coordinator's, so you can get past that person's talent, time, and attention being tied up and have the menial under control.
Jeff: In a dental practice, if someone is managing correctly, one of the immediate menial tasks is usually data analysis. Analyze the AR, for example. Software is only going to do it based on whatever queries I can use, and not every software is the same. Some don't have queries. Eaglesoft is my least favorite dental software. I probably shouldn't say that, but I will. If there were some way to use AI to analyze that information so that, when I'm trying to sort out my accounts receivable, I could ask, "Where are my problems?" and in 30 seconds it gives me back the people I should contact immediately, that saves me hours. That's the thing I enjoy about it: you can really save a lot of time.
Dan: If people just get a concept of what would be advantageous to them, even if they don't like AI or aren't good with computers, make that list. With this whole new industry of agentic AI coming, there will be agentic AI specialists who will come in and build that for you. You don't need to figure it all out, but you need to know what it is you do, what you want, and what could be alleviated. So make a list now of repetitive, menial administrative tasks. Not taking out the trash, because we're not at the robot stage yet.
I don't know if you've seen this, and I forget the website, but now AI agents can rent people. There's a website where people enlist themselves with their name, their rate, and what they specialize in. An AI agent will contact you and hire you. One agent wanted someone to try 10 different spaghetti sauces on the same exact pasta: go buy the pasta, go cook it, here are the instructions. The agent came up with this, found the person, hired the person, and did the transfer. Obviously it was authorized by some company, but the AI agent did all the work. It hired however many people to do the same thing, tabulated the results to figure out which spaghetti sauce was best, and reported that back.
Jeff: See, us humans still have use. We can taste spaghetti sauce.
Dan: It's kind of Uber Hands.
Jeff: Uber Mouth. Oh, my gosh. All right, so that's where we're going. Now what do you think is just hype? Any time something goes into the media, you get murder hornets and the whole nine yards. What, in your view, is hype and not very real?
Dan: I would stay away from anything that says all of your marketing can be handled by AI now. That's being promoted. We know because we see it, and we've even done dummy trials of some of these things. It's not just not good and a bit nasty. It's actually dangerous. It promotes things practices shouldn't promote, like things that violate the dental practice act of their state, and it promotes services the practice doesn't even offer. We're just not there yet. It's good to think we're going there, but these companies should not be promoting that yet. We're not at the point where AI can handle marketing from start to finish without human interaction.
Jeff: Got it. Anything else that's hype, with relation to all this?
Dan: I like the note-taking tools, but I would caution any practice to really challenge the HIPAA compliance. HIPAA compliance needs to cover not just the platform hosting the AI agent, but also the AI provider it's sourcing from, such as Gemini or OpenAI. I've seen companies say, "We are HIPAA compliant," but where they're accessing AI from is not, and they're passing information through to it to get summaries and notes back. The alternative is having your own version, which is very expensive, to have your own AI in-house. So be very thorough in checking into some of these things. In the healthcare industry you need to think about that. In other industries, not so much. Spaghetti, you don't need to worry about that one.
Jeff: That's true. All right, so we covered a lot, which I really love. I'm very excited about where this is going. We all love AI. So if I'm an owner, doctor, or office manager listening right now, and we're not doing any of this, where would you recommend I start?
Dan: I would definitely do an AI after-hours receptionist. You'll pick up two new patients a month that you otherwise wouldn't. Put it in place and get your team grouped in on it. You might hear some blowback: "I don't like talking to AI." Maybe one out of 100 patients will say that. And it's after hours, as opposed to voicemail. The same person probably hates voicemail. Start working with that. If you're doing ads, that's a great time to advertise, at one third to one fourth of the cost. See what you pick up from that.
Then start moving into the next realm, which is learning AI to analyze where you sit with AI. Start playing around in Gemini, Grok, and ChatGPT. Ask, "What's the best dental practice around here for dental implants?" Or Invisalign, or clear aligner therapy, whatever you're looking for. Ask why they recommend those practices. Tell it your practice is ABC Family Dentistry and ask what you're missing. Use that as a shortlist. Get that to your marketing company and see if they can start to scratch that itch for AI. If they can't, find another company. Don't abandon your existing marketing that's working, but find a company that can help with that, or start doing those things yourself. "This practice over here has seven testimonials on their website, and you don't have any." That's a little bit of a battle plan you can put in place for your practice.
Jeff: That is very cool. Dan, thank you so much for being here. This is awesome. I'll put your link to Blumberg Digital so people can find out more. Do you have a special page you want them to go to?
Dan: The main website is fine.
Jeff: We'll put your email address up there as well if people want to contact you. And if you want to see more of Dan, he's one of our main speakers at the MGE New Patient Workshop. You can obviously see he knows his stuff when it comes to marketing, so I'll put a link to that on the episode webpage as well. Folks, Dan, thanks for being here.
Dan: My pleasure.
Jeff: There you have it, folks. We do hope this helps you sort through the hype and see what you can use now to improve things in your practice. If you want to find out more about MGE, you can call us at (800) 640-1140 or find us online at mgeonline.com. Have a great week, and we'll see you at the next episode.