AI in Dentistry: Front Desk Automation with FOJI.io Founder Mitch Hersche

Welcome everybody to another episode of the Dental Boardroom podcast. I'm continuing my series on AI in dentistry and today I have a guest on the show for one of the, I think the leading technologies out there using AI as adoption of of of IT into dental practices. So I'm excited to go over, over, over some really good questions about this technology with the founder and owner of.

Before I do, I just want to emphasize why I'm doing this AI in Dentistry podcast series. We've now done a couple episodes. I believe that this is an area that we need to stay ahead of it as competitive dentists, which my clients are. I tend to work with larger dental offices as a more of a cfo, highly involved service to stay competitive. When new disruptive technology emerges, we, we can't ignore it. Now at the end of the day, a great dentist is going to provide great clinical analytics and treatment planning. However, if they can support that human judgment and their skill set that has taken years to develop with really great technology, I think it will help them increase profitability and stay ahead of the curve in what is going to be undoubtedly a very disrupted, disrupting technology within every industry and definitely inside of dentistry. So thanks everybody for joining in on this series. AI in dentistry. Let me now jump into our guest. Our guest is Mitch Hershey. Mitch is in Utah. He started a company called Fogi IO and it's all about AI application into dental offices. Now this is going to be, I believe, a little bit more on the operational side and perhaps a little bit less on the clinical side. But I will have him clarify specifically what does his technology do for dental offices. Graduate of University of Utah. You've been in different positions as a business analyst and product manager. And now in 2021, I believe the founder of Fogi AI. Mitch, welcome to the Dental Boardroom podcast.

Thank you, thank you. I'm so happy to be here and talk about my favorite subjects, dental and AI.

Well, let's launch into it. What got you into each of those two and was it simultaneously when you did not simultaneously.

When I started this journey into the AI path, it was all because I became a product manager over the AI and analytics team at a larger tech company here locally. I was also in charge of what's called a DevOps or Development Operations team at the same company. And quickly as a being over the product, I learned, you know, there is nothing really to help develop and launch AI products quickly. And so at that time I was like, oh, let's jump into this world of AI, see if we can't build something to help companies release AI faster into their product and into their organizations. So that's actually where I started with that when it came to dental. I have a long family of healthcare. My grandpa was a doctor, my father was a doctor, my uncles are doctors, my mom is a dental office manager, my aunt is a dental office manager. And so eventually where these two things came together is my mom and my aunt asking me for some help with AI doing things in the practice. And so I was going a whole different journey with this AI thing, only to be sideswiped by my mom who decides that, hey, I need help with this stuff. Can AI help? We build it, we find traction, and what do you know, AI is a perfect fit in the dental practice. So that's kind of how we ended up bringing those two things together in my life journey, at least.

Now I have a curiosity driven question here. You've got an MBA and then you've been in as a, you played the role of a business analyst, a product manager. So you've very much been on the side of business and, and the delivery of, of, of a product, not necessarily in the coding or the development or the manufacturing of a product. So how have you brought in the tech side? Do you have a development team member? Have you learned it on your own? Because there's definitely a development element to this, right?

Well, I'm not going to pretend like I'm coding anything. Great. I have a co founder, his name is Chris Hansen and he's, he's wicked brilliant at what he does. I do have a background in developing tech, especially on the business side as like a product manager. Little secret. I did go get a business degree at the University of Utah. But also while I was there, I was a dual master's degree and I was in the video game program up there. It's a pretty good program.

I didn't finish that program. I'm a, I guess I, I had to drop out of something if I was going to go into tech. And that turned out to be the one. But that taught me how, how to develop software. Because developing video games is the same as developing software. So that got my journey started into tech. First making video games on my master's program, which then spun me into developing a product for tech companies.

Are you looking to buy or sell a dental practice? If you're a seller, how do you find a strong list of potential buyers? There's no MLS or Zillow for dental practice sales. In such a fragmented market, many dentists selling their practice feel discouraged. That's why I built practiceOrbit.com practice orbit is modernizing how dental practices are bought and sold. It's a platform technology. Y' all know them Zillow, Airbnb, Uber, and OpenTable, just to name a few. These technologies do two things. First, they help match buyers and sellers by centralizing the marketplace. Second. Second, once the match between buyer and seller is made, these technologies help them navigate the sale to a successful close. Here are a few of the features inside a Practice Orbit users can determine a reasonable sale price for a practice in just minutes. All they need is the most recent business tax return. Practice owners can use the directory to find important service providers like practice brokers, attorneys, accountants and bankers. Practice owners and their brokers can invite those service providers into the platform. So all parties are coordinating the sale together. Practice owners will have a beautiful digital listing of their practice, like a home on Redfin. And importantly, the listing will be anonymous and personal information like your address, name and pictures are only released after an NDA is signed and approved by you or your broker. Practice buyers can create saved searches to be auto notified when new listings come on the platform that fit their desired specialty and location. And hey, dental practice brokers, you can list all your practices on the site for free. By doing so, you'll get access to Practice Orbit's large pool of buyers, all of whom have created profiles and saved searches to be auto notified when new listings come for sale. You can use the platform to introduce your listings to a targeted set of buyers. Additionally and optionally, you can use the platform tools such as a built in NDA, a customizable loi, and a workflow tool to navigate the sale from LOI to closing. In fact, you can manage all your listings on the system, from listing the practice to NDA, to loi, and finally to closing the deal. Create your free account today at www.practiceorbit.com.

So mom mom's practice or that dental practice gave you a sandbox to go and explore, discover trial and error, beta test. And you developed some tools that were very applicable in a very specific environment known as a dental office. And you said, hey, I think that other dental offices can benefit from that. And hence you launched this business model around it and expanded it, built a team marketing everything that starting a business entails. So that's the origination story. Now let's dive into Given you've been heavily focused on AI every single day in this space now for four or five years or so, which is kind of when most of the world started to at least learn about AI. For most of us it's been a more of a peripheral thing that we hear about. A lot of US are using ChatGPT or any of the AI engines just as a, as a better version of Google for getting information. I think we're all pretty familiar with that. We're pretty familiar with having it sort of draft up some narratives for a document, at least as an initial draft. What most people are not familiar with is how exactly is it going to be used to create automations in a business? How will it replace human intelligence? Because that's what AI is. It's not just a automation tool. It's an automation tool that thinks critically, which makes it, which makes it different and more powerful if used correctly inside of a business system or process. So my, my first sort of general question for you is where is AI already making an impact in dental offices today? And how do you think that evolving over the near future, say two to three years?

Yeah. So interestingly enough, I think the first place that dentists have been seeing real practical use cases for AI is not what you would even expect with like Chat GPT. There's a great company here locally in Lehigh, where I'm located, called Pearl AI and Pearl AI.

I've got them lined up for an interview by the way, but go ahead.

Perfect. Well, Pearl AI does a great job at feeding a bunch of images from the, from X rays into a machine learning model which then has been trained to look for cavities and other issues in the teeth. I think that that's done a really great job of getting at least the dental practices a little their toes in the water to go, oh, I can trust this stuff. This stuff is going really well. And I think that that's what you're seeing first from these AI companies that is allowed, that is in a dental practice actually creating value. I think what's next and what we're seeing next in the next three to five years is eliminating those administrative work. Not necessarily the deep thinking as much as, you know, the RCM process. You know those jobs that maybe have lots of turnover, that are mundane and administrative and they're just something that's point, click, point, click, doing it over and over and over again. Those are the next things that we're going to start seeing AI step in, especially agentic AI start replacing those jobs and creating more and more value for in the next three to five years.

Yeah, if I had to divide up AI into two broad categories, it would be the clinical side, which is the diagnosis Treatment planning, risk predictions, things like that. And then there's the operational side, which is how do you create efficiencies within the processes around the operatory room. RCM being a really great example, and we're going to talk about this with fogi, is verification claims, dealing with denied EOBs and, and how to solve those all faster and ideally keep your AR a lot lower, which just means that you're collecting much more effectively from the insurance companies and patients. So, excited to talk about what do you. Let's actually start to move into the AI agents here. I think AI agents is something that we are all trying to conceptualize in our mind, practically speaking, what an AI agent does. Now, what most of us are familiar with is your ChatGPT or Grok or any one of the big AI models out there. We know how to use it as a prompt and a retrieval of information. And we're increasingly becoming data retrieval machines and we're becoming much better at it. We're learning how to structure our prompts to get data more accurately. Everything, every time I do it, the response that comes out of ChatGPT, which is the one I use, I use a paid $20 a month subscription, is just so phenomenal. Not perfect, but it is phenomenal. And I'll always rewire it a little bit to be more specific to the way I view that particular thing. But we all understand that the AI agents I want to talk about a little bit. How do AI agents or software running AI agents differ from traditional SaaS? A SaaS is a website, it's a software, that's a website, it's called software as a service. And if you go back, I'm going to do some basic software 101 terminology here. If you go back 25 years, most software was installed on the desktop and you would get a CD and you would download it and you would open it up from your system's tray or the bottom of your computer. Now, most of those softwares are online and you access them through a URL, a website, and they have the front end, which is just a static website, and then the back end which is a software. And that's what SaaS is. And so the question is, how do AI agents or softwares running AI agents differ from these traditional SaaS tools that have been used for things like automated client communication, for example, or submitting EOBs, or even some of the practice management softwares going online, like Fuse for, I think Eaglesoft and Dentrix. Ascend Curve is another one. A lot of them are now going SaaS as opposed to desktop. And let's actually preface that question with what is an AI agent? Number one, give us a definition of an AI agent. And number two, how are AI agents being used in lieu of or in conjunction with traditional web SaaS softwares?

Yeah, so let me start with the first. An AI agent. There's, there's really two ways of, to think of an AI agent. An AI agent is first a workflow. So if you can imagine a workflow of I log in, I go, and I push this button, then I fill in information here, then I hit this button to submit it, etc. Any kind of workflow you can think of, an AI agent is something that actually performs those workflows. And then at each point in the, in the workflow is actually performing some kind of task. Right. There's also the AI agent that just has access to your data or information. And you can think of this more like a chatbot where it can communicate with the customer, just automatically make decisions on its own, and then be able to actually make a decision or actually perform a task based upon what it's getting back from the other user or the customer. So as an example, AI agent reaches customer, decides to text message dental practice and says, hey, I need to change my appointment. Well, instead of having a human on the other end, you have an AI agent that is able to communicate with that customer, go look at the actual schedule, find a time that would work, and actually set that appointment for that customer without having to have a human interject any kind of information that way.

Do you think it's safe to say that dentists in general are not going to be creating their own AI agents, their own AI kind of chatbots, and then integrating them into their process, but rather instead they're going to look to companies like yours that have already done that work and integrated those chat bots, those AI agents, with existing tech in the tech stack of that dental office to create the workflow. Safe to say that's going to be the route most dentists are going to go with these AI agents, correct?

Oh, 100%. I always like to describe AI agents or AI in general, the way it's going, the way it's, it differentiates itself from traditional SAS is old SAS tools are like, I gave you a shovel to dig a hole and you use the tool to dig the shovel. An AI agent is literally like, here's a robot that is going to dig the hole for you. So not only is it going to take the labor and the cost of the labor. It's also going to replace the tool it was using previously. So that's why this is a such a huge step forward is we're no longer just handing tools to people and asking people to learn how to use the tool, be efficient at the tool, and figure out how to dig the hole still on their own. We're literally saying, here's a robot who also has a shovel and is now going to dig the. Dig the hole for you.

So I'm going to keep this conceptual to start. And then when we jump into your software or your, your AI tools, I want to talk about specifically how they play that role of actually doing the digging for you, but sticking on the concept here a little bit longer. If I think of a process, every business has processes, a practice cfo, I have a million processes. Getting financial statements timely out the door every month, filing a tax return, preparing a cash flow forecast, doing a tax analysis. There's processes that you pull from different softwares and different data points. And then at different stages in the process, the computer and the software is doing something. And at different stages, there's a human brain interjecting in that milestone or in that marker in that process to do some thinking. And then it hands off the thinking and instructs that next step that may go back into the software or the automation part. So it's this mix, this journey in this process is a mix of automation and human thought. What AI is doing is.

It'S replacing some of that human thought. And right now it's not all of it. And we don't know, I think right now to what extent it will fully replace human thought during that process. But think of it that as a, as the process goes through each of its stages, when it replaces what normally has been a human for an AI agent who is thinking like a human and in many cases will think a lot better and a lot faster than a human. You have to place in that step of the process, in that place of the flow, this AI agent, where it then dips into the language, the large language model, dips into the data inside that, whether that's a General LLM like ChatGPT, and it extracts it out of that. Or you can create your own data that's housed inside of a chatbot. If you want very, very custom data. It'll be a lot less, of course, than going out to like a ChatGPT, but at least you can pull from maybe all of your clinical notes or trainings or processes or sop, statement of procedures and it pulls in and it builds its own. So it's a very custom mind that's doing that human replacement section of that workflow. But in concept, that's how the AI works, correct?

Yeah, I think that's a pretty, pretty good way of describing it.

I'm a novice here doing my best to explain how I'm in visual, sort of visualizing the way AI works.

I mean the ticket into two things like a ChatGPT is going to be very broad, it's going to be very trained up on the Internet, things like that. And then you're going to get the custom ones. I mean a little self plug here, but like a fogy where you're putting your data into fogy and then you have a large language model that can, can actually use your data to then answer questions based upon specifically your practice or your business. So yeah, so I'm going to go.

Out of order here a little bit on my questions because I want to ask about that.

I can't do that. I'm incapable.

Just do we need to replace you out with a chatbot here?

Let me go get Foji. It'll finish up.

All right, so here's my question related to this. You have these LLMs, which are these bodies of data that are used to do the retrieving and the thinking and then.

The result out of that kind of prompt. Are you using the general LLMs out of there? Is Fogi dipping into the general LLMs or are you creating your own data? I'm not sure the exact word, but the data kind of warehouses that you're getting from all of your clients. Are you sharing data across dental offices in order to create a bigger brain, so to speak, for AI agents that are built into your, into your, your agentic process, these agent based processes. You follow me on that question. Do you ever feel lost in your practice finances? And how about your personal finances? Ever wonder where your hard earned money goes and why you're not seeing the financial progress you work so hard for? If so, listen up. As a dental CPA and financial planner, I've created our company Practice CFO just for you. With almost two decades of experience, Practice CFO integrates great CPA services like accounting and tax with customized financial planning, both business and personal. Think of it this way. Our dental focused CPA services give us the X rays we need to clinically plan your global finances and be effective financial planners. It's a highly customized approach we call the CFO model. Historically, having a CFO or Chief Financial Officer was an Advantage of large corporations only. But today, practice CFO gives that same advantage to private dental practice owners. As a client of practice cfo, you'll be assigned one of our dedicated CFO advisors. You'll receive insightful financial reports each month, have regular financial review meetings, reduce taxes, grow your wealth and feel more in charge of your numbers. Because we believe that in the competitive landscape of dental practice ownership today, you have to be an effective business owner, not just a capable clinician. Reach out to learn more at www.practicecfo.com.

Yeah, I hear you. We think about how to say this. We definitely never mix our customers data, right? It never all goes into one place to help train any kind of GPT or anything like that. We do rely on the big models, ChatGPT, Grok, etc to do the large language processing. Everyone who's doing any kind of generative AI outside of those companies is going to have to use, utilize one of them. It's too expensive to run your own. It's billions and billions of dollars. Right? You're going to train a model like that would just, it just, it's astronomical in price. But what you do do is you do take their GPTs or their models and, and you put them into your environment and you run them. Now what that does is if you're worried about HIPAA compliance or anything like that, if you're running it within say an environment like fogi, you're just using their models to essentially do the writing of the language. But it never gets pushed back out to the Internet. It stays within the software and, and then it sits on top of your, like you say, data warehouse for each customer. And then it's able to do everything that a chat GPT would do with the, the data of the individual practice or practices, depending if it's a DSO or an individual practice, all within one model. So then you can get the power of a chatgpt or a GROK or a quad without the, the security issues of putting your data into the Internet. And it can be trained up on your own data. So it's kind of a weird mixture of all of it.

So let me see if I can rephrase that to confirm understanding. FOGI will pull from the massive data warehouses out there, which really is just the universe of the Internet that ChatGPT or whatever LLM is being used is pulling from. Because that's the, that's where ChatGPT pulls from is just everywhere that's publicly on the Internet. Correct. And there's a ton of clinical information out there. And other things and you pull it and that's a one way direction. It's going into fogi. FOGI is never taking information that it collects from dental offices and sending it back out. So that's a one way street of data flow from the big LLMs, large language models like ChatGPT into FOGI. But FOGI then will have input directly from the user, your client, and it will mix the two into one data center which then the AI can use to produce a result, a set of data, an analysis, a write up or to do an automated task like communicating out to a given patient. Now that second is a two way street. Data's coming in from the dental office into this, this language model. And then the language model is also sending data back to the dental softwares that are in, in that workflow. Did I understand that right? Perfect.

Exactly perfect. That's, that's why it allows things like, you know, if you wanted to have your own AI to be able to say how many patients did I see yesterday? And you can just ask that question. It can go and look at your data from your practice and then go, and then it uses the power of the GPT model to actually answer the question and write it out in a, in, you know, the language that's best suited for you.

Okay, a minute ago, going back to definitions, you use the term generative. Generative AI.

Yeah, generative AI.

Explain that and compare that to other versions or other formats of AI.

Yeah, so before now, today everyone just hears AI and they just think chat GPT. Well, before chat GPT there was a lot of AI or what we'd call machine learning models that were really cool, and they're still really cool, but they're more of what we would call a predictive model. Right. These are your decision tree models, your regression models, but these are all designed to make a prediction. Right. So if it was something where I want to use an AI model in order to predict what I think my revenue is going to be in my practice next month, I would actually use a predictive machine learning model, like a regression model to do more data centric AI predictions. Generative AI on the other hand is going to be things like generating images, generating language, the things that we're all used to and accustomed to now as AI, from chat GPT to Google Image, all of these things, that's more on the generative AI side. So you kind of, you get, you kind of get these two worlds where one is very much more data centric and it's going to be do more predictions and then the other one is going to be much more like we talked about, generative, more, maybe more creative on the language and imaging.

Got it. Okay, that's a good, good clarification. Any comments on this going, sticking with definitions just a little bit longer. Neural networks from what I read, AI is really built on this concept of neural networks which in a way is, is paralleling how our minds work. Now this is very, this is a very tech term right here. I'm not sure if, if you feel confident really addressing what these neural networks are. You want to take a stab at that?

Yeah, I can. I know all about neural networks. So the difference between a traditional machine learning, right. Machine learning, the weights of and the parameters of the model are still defined by a human. A human will say I want to wait. If I'm trying to predict how much revenue I'm going to make, maybe I'm going to make, put more importance on the weather and other things in my practice to give me a better prediction. A neural network is actually going to do its own weights. So it's going to go back and it's going to look and it's going to make decisions on its own on which data or other things were actually giving us better scores. And so then it begins to push all of these together and all these different decisions and all this different data. And it goes back and it's looking back at the data to make the best decisions for you. And so then the AI is actually creating a full on neural network on how to make decisions while making its own decisions on what's, what's most and least important in the decision making process.

It's almost like the neural networks can they look at in very, in a lot of detail at patterns at relationships as opposed to just retrieving and patching language together.

And I hear about these tokens that I went to an AI training all day training course one day and they were talking about how they're, there's these clusters of.

Tokens and it's all about pattern recognition in these AI model the way they think. And so it's not just retrieving data, going out to websites saying oh this looks like it's related to the prompt and just sort of taping these different narratives together from different sites. It's actually completely creating a brand new original explanation by looking at patterns of languages and these tokens and recreating a response. And I don't fully understand it, but it's very different than just data retrieval.

Well, it's wild like, even when it's writing out, it doesn't just memorize the word, you know, words. It's literally doing a probability on what's the next letter that should be used. Or if I use, you know, if I start a sentence with I, the probability of the next word is highest at this, related to this, this, this, and this. And it makes a decision to say was right. And so it all comes back to essentially probabilities and that decision making essentially is what they're calling it. And just using data to determine what even the next letter, word, topic, everything that it's going to go for. And it's just these GPT models are all just stacked on top of each other to make it just go super fast.

And one of the things that I'm trying to assess, I think a lot of people are trying to assess, is that as powerful as these AI models are, unbelievably powerful. They make very basic mistakes. And as you know, these are called hallucinations. Like, give me an image of. I saw one of a, of men in a wood workshop carving a bed, you know, or building a bed. And it showed these men, and one of them had three arms and one of them had six fingers on it. It was completely original and it was a beautiful image. But there were these hallucinations that were hilarious. Or if you say, give me five synonyms with the word excited that are all five letters, you'll end up getting some that are like four letters. Just very simple. Mistakes like that from these hallucinations caused some of us to be a little bit hesitant to saying, okay, how much can I really trust this? If this is going to replace a human to think through a decision and produce an output, there may not be a review of that output. There may not be a proper validation before it goes out to the patient or the client or the end user of that. So can I fully trust it knowing that it does make these errors known as hallucinations? And of course, over time, it'll get better and better as it pulls from data and self learns and possibly eventually get into a, a gi, artificial general intelligence, where it essentially has a consciousness of its own, then it can sort of think through some of these more. But. All right, let's now dive into fogi. You ready?

Okay, let's do it.

All right, Fogi. So how does fogi actually work in a dental office from the moment a patient is scheduled to when a claim is processed?

Okay, so let's say patient schedules an appointment tomorrow. Well, let's even say same Day, same day, patient gets scheduled immediately. FOG as an AI will notice that that has been scheduled. It will then grab the, the data from that patient, their, you know, what insurance carrier they have, what's their member id, et cetera. And FOGI will actually go to the portal, the insurance portal. It will go log into the insurance portal and it will then go and search that patient. It will find their insurance benefits. It'll notice if they're still active insurance or not. If it is not active, it'll immediately send a text message in the office to let them know, hey, this insurance is no longer active. Make sure you're aware of it when they come in. If it is active, it will log that right back into the PMS system. Then it always checks, and it depends on the customer, but a certain customer can go, hey, I want to know if that patient hasn't visited the office within the month. I want it to go and look at all the benefit information. And then I want you to put it back into the PMS system for us. And so it will grab all of their benefits, their insurance, their eligibility limitations, and it will write it back into the actual PMS system for the practice. From there, it can, they go ahead. They can go and do you know, they're verified. They go, they get procedures done. Once that's done and the doctor notes come in, everything is done. FOJI will then trigger again and go, okay, it's all done. I'm going to go and create all the claims. I'm going to attach the X rays, I'm going to write the narratives for the claim. I'm going to make sure all the dental codes and everything are accurate for the eob, right? And then it'll batch it all together and then someone in the office will go and review it and then hit submit.

And then if there's ever any denials or appeals, those come back fog. You actually read why it was denied or appealed, recreate the claim, and then have it resubmitted automatically. And so the goal is to actually have what you would call, you know, have full automation across the entire RCM process to eliminate that administrative work from the time a patient walks in to the time they leave. You're not necessarily having to deal with the mundane work of creating those, those claims and dealing with insurance. Because, I mean, I know the problem. The biggest problem we have is when we talk to practice, they go, my gosh, we love working with the insurance companies. Why would you, why would you stop this? We don't want to talk. We Want to keep talking to them. We love sitting on hold for two hours to understand the limitations on this stuff.

I sensed some sarcasm in that as soon as you started that sentence.

Purely sarcastic. That's the stuff that we want to get rid of is the, is the obvious of we don't want to do that anymore. We don't want to work with insurance companies, we want to just submit our claims and get paid. And we want to get rid of the administrative work that has to go along with that. And AI agents are perfectly positioned to take over that kind of administrative work.

So question on the implementation, because I hear that concept and I'm like, wow, who would not do this? What I've learned with technology, even just traditional software, is you, you get the demo and you're like, that's so impressive, let's go. You sign up, you start your subscription, since most things are subscription these days, and then you implement it. And two thoughts. One, the effectiveness of it is determined by how well it's implemented and adopted by your team. Everyone following the use protocol of that software and the training on the software. I have this belief that softwares are very deep and layered and on average people are using 30 to 40% of a software. Like we use NetSuite for our accounting here we're larger. QuickBooks was just not big enough for our needs here at practice CFO. And NetSuite is so layered we're probably only using 20% of what NetSuite has to offer. And I suspect there's a lot of dental softwares like that as well. And so the first key to success is I think really effective proper adoption and training of the team on how to use it. Because even though we'll create automations, I suspect it's not plug and play. I suspect that there's some initial configurations and training so that people know what are they not doing now and maybe what are they now doing that they weren't before. So that's my first thought. Second thought is on getting demos of software companies. I often find that the elevator pitch isn't quite as effective as the pitch presented itself. And I, I actually have a fairly high tolerance for software evolution as you know, as a sot, as a, as, as a tech person, that software development goes through these, these iterations. It's called agile development. And you first get the skateboard and you use it, you get feedback, then you put a handlebar on it, you call it a scooter and then you eventually put a motor on it and you got a little moped you know, and you keep going until you have a car, a truck and eventually this jet, you know, and, and so fogies are relatively new software. Out of curiosity, where would you say it is in that agile development process? In other words, how many minimum viable product stages have you gone through in your own iterations? As a tech company?

We are actively in dental practices today. So this isn't just like a hope or dream, this is something that's actively working today in dental practices. So we have dental practices right now and I guarantee you right now there's a fogi AI agent running that's looking up some dental insurance plan on one of these, one of these portals as we speak and it is actively writing into these PMS systems and it is working very well. Now back to the point of the training piece. It's a little bit different than traditional software and this is where a lot of change on expectations is going to be needed because traditional software was, oh, I need to learn how to use this and then I need to use it every day. Our hope is it's almost like getting a new employee. You need to train that employee and then one day you just hope the employee does their job right and you don't need to hold their hand every day and know how and show them how to do everything exactly the same. With the AI agent, we spend more time up front making sure that we understand how the workflows of the practice work, how they want certain things written in certain notes, how they want things logged in. In some cases it's like, hey, we can get you all 300 dental codes and how the insurance actually pays those, or we can get you and get the 40 that actually are used in the practice so it doesn't get all muddied up. It all depends on the practice. But we train it up front and then the goal should almost be that you don't even know it's going. You don't log into fogy. You wake up in the morning, you look at your PMS and you go, oh, insurance is verified, their benefits are all right there and my goodness, our estimates are near perfect.

And is it an accurate view or understanding that as the AI is used in the practice, it will learn upon itself so that the training becomes less and less just like an employee. You provide that training and they start to understand the big picture, see the forest through the trees, start to make more judgments on their own and eventually they're sort of self sustained. And you're saying that similar with AI. The beauty of AI is that it's going to be a lot less expensive once it's sort of running its itself.

Yes. And the other thing about it too is you have to remember some, I mean no one listening to this is going to, is going to think this and it's not true. Everyone listening to this is probably runs perfect practices. But, but there are some practices that are better than others at running these operations. And our AI is going to, we are going to, as the developers of the AI, we're going to find the best practices from the best practices to implement into the AI. So that even though if we go into a practice who may be not the best practice, it's going to be forced to essentially adopt the best practices from the other practices. Does that make sense?

Interesting. Yeah, that does. Interesting. In other words, your software fogy is learning the best way to do things among its users, among your clients, your users. And then as it does that, that training can be adopted or that training is now benefiting new users coming onto the to the platform without a disclosure of sensitive information from other users of the system.

Yeah, because it's more of us training our own AI agent. Right. So yes, the AI agent gets trained from the practice itself, but then we still are as the developers of the AI agent, train that AI agent ourselves. And as we're training it from best practices, it's going to take those things from practice to practice. And once we've seen some that are like, oh, brutal but, and they try to tell us like, no, no, no, this is the way to do it. I'm like, look, and I'm telling you the AI's got this pretty figured out. You're probably going to want to let it do its thing. And once it does, all of a sudden they see their denials start to drop considerably because the AI agent tends to know a little bit about what it's doing.

So I want to talk a little bit more about the actual onboarding or adoption of the software. Do you have a dedicated implementer of the technology on your team? How much of it is just off the shelf, plug and play. And do you think doctors can self implement the technology?

No, I don't think it's you. We, you will need what we use as like a customer support and an onboarding and a customer success. And they will be very, very much, like I said, hands on in the beginning and hopefully get things done in the beginning. But no, it would be very difficult I think for a doctor to just pick up and do on their own. You're going to want to like implement, you know, because how it works from one insurance provider to the next insurance provider is going to be completely different. And you're most likely going to want to make sure that you have someone that's getting it set up for the way that you're practiced, especially if you're pediatrics versus a general dentistry, things like that. You're going to want to make sure that someone's there to make sure that AI agent is aware of what it's looking for and it's going to, you know, do the best it can for your practice.

Yeah, I highly recommend doctors don't DIY this, get an implementer. It may be a little bit of a cost to get it implemented, but these are the kind of things, if you configure them right at the onset and you start the sort of the initial journey of the adoption the right way in the right direction, your chances of success go way up. If you don't, people get frustrated, you drop it. You're like, that isn't what I thought it was. Well, it was actually user issue more than it was a software issue. So a big advocate of, of, of getting an implementer. Now, how much of your software has to be one time configured to integrate with the existing tech stack and how much of it literally is it's already integrated with Eaglesoft or Dentrix or other softwares out there that it, it that that part is just automatic.

So that part's automatic. We already have those integrations. So when you, if you're a dentist that you came and you wanted to start with Fogee, one of our first calls is just to get set up with the integration directly to the PMS system and then it just starts grabbing all the data from it. And then we have these AI agents already built up and ready to do things with the data. And so it's, you know, connect the data, get it flowing in, spin up the agent, change it a little bit to how it like, you know, add some extra training to those agents, let them know this is what it's going to need to do for this or this or this is. These are the notes we want to put in this area of our PMS and then turn it on and then get it and let it start going.

Got it. Okay, what practice management softwares which is probably the key integration do you, are you API in with and which ones are you not?

Open Dental, Dentrix, Eaglesoft, we're all with those and we are currently working towards Care Stack and Curve.

Okay, Care Stack going, Care Stack going a little Bit more toward the DSOs. Yep, got it. Yeah. I imagine the DSOs, which are heavy, focused on business processes and operations, are probably going to be some of the biggest and earliest adopters of AI. Are you seeing that in your new users?

Yep. DSOs, they get it right away because they see numbers at scale. Right? And when they see numbers at scale and they see this AI agent that's able to, you know, get more accurate numbers from the portals, get more accurate estimates, able to reduce labor, et cetera, they just see, they see that number at scale and what it's be able to do for them and they're willing quickly, like if that's, if that's the case, let's get in and try it. Or an individual office you might see a little less like, well, we really like Judy and Judy is really great for us and we're happy to keep Judy looking at the portals and making the phone calls. That's fine, I think that's great. But yes, DSOs for sure are thinking more at scale. How can we automate faster and do more?

Well, because the DSOs are saying if we implement an AI process and agentic process. And by the way, when I say agentic process listeners, I mean a process that involves, involves AI agents and we implement these agentic processes, we're going to be spending 15% of what we were paying Judy to do that, to do that work. Now it sounds a little, I don't know, unsympathetic to say, oh, we're going to replace Judy out because it's going to save us 85% of the cost. But there's no avoiding where this is going, there's just no avoiding it. And Judy's either going to have to learn how to use the AI to do her job better and faster so she can do other more valuable things in the practice, or Judy's probably going to be out of a job in time, or your practice is going to be seeing decline collections and possibly run out of business. And I don't know if it's like a one or two year threat that AI really will distinguish those who have adopted it and those who haven't in dental practices. And there's already a growing divide between operational efficiency at the DSO level and the private practice level. And I think AI could expand that gap is a concern for me. I'm very much a proponent of the private practice space here at practice CFO and a lot of the benefits offered to private practice ownership. And this is why I do think that dentists can't ignore this now, if you've, if you've got a few more years before you retire, yeah, you probably don't need to go completely restructure your operations to make this work. Maybe, maybe not. But anywhere else on the professional life cycle, you got to be on this. And I don't think it has to be as hard as intimidating at sometimes we think with these complex AI concepts. And so companies like FOGI make it much more easily accessible to bring in to your dental office if you adopt it correctly. And it may be that your costs actually go up in the beginning for the first three to six months as you adopt it, you're shaking your head like, no way. No, Wes, I'm saving money right on day one.

The only thing I would say is like with AI, you'll want to be just a little bit more patient. I think with SaaS, people were expecting more turnkey. Give me the tool, get it going and go with AI Again. Think of it like an employee. It's something you want to be a little patient in the beginning. You want to make sure that it's doing the right things for you. But once it gets going, there's so many opportunities with it, from labor costs to your accounts receivable shrinking, less denials, your estimates being more correct, all of it comes back together to just save that practice time and money.

And I want my private practice owners and really all my dental listeners to remember that the key value that you're providing to your patients and the public is great dentistry, a great clinical outcome. So they walk out of your practice, they like their smile better, their teeth are whiter, they have less pain, they have better oral, systemic health, all of that, everything is peripheral to that everything. And as much as we may be sympathetic to maybe a front office person who's going to have some of their job replaced and maybe even have to replace some actual employees or downsize. If you do fully embrace and adopt AI, at the end of the day, you exist for patience and AI is going to create new jobs for people and some people are going to lose some jobs and you're not less caring or sympathetic if you embrace AI, if at the end of the day it's helping you drive more clinical, better outcomes for your patients, that's the way I would really look at the human side of it. It's to empower the dentist. I don't foresee it ever be to replace, replace the dentist, but it can empower them in so many ways. And working with so many dentists on the financial side I have seen over the past 15, 20 years since 2009, so about about 27 years since I've got in or 17 years since I've got into dental, that the there is compression on profits. And that is from institutional dentistry, PE business people really getting the suits, we call them, getting into dental offices. And also the lowered reimbursements from dental insurances as well is compressing profits. And this is an opportunity for early adopters of AI to regain, to recover what I think were the appropriate amount of profits that dentists used to have before all of the pressures on them. I mean, all the years of school, all the debt, all, all the risk taking on a practice, running a business, they absolutely 100% deserve to have that. All right, let's pivot now back over to Fogi directly. How is Fogi different from other revenue cycle management tools dealing with insurance verification, claims processing, denials, trying to reduce our AR aging schedule? How is Fogi different from other RCM tools and third party billing services?

Yeah, so I mean, I'm going to sound a little bit like a repeated record here, but it's literally doing that labor piece. Right? So the other RCM tools, they're probably really great at showing you the data. They're really great at maybe, you know, helping you calculate something, but it still requires that person to go in, take that data, maybe move it over here, do your own calculation there, and have that person do everything. Where FOGI and where AI agents entirely are going to be different is going to be reducing the actual needed labor to do those kinds of mundane tasks.

Got it. Now you have various products on your product page of your website. You have AI agents, you have Ask Fogi, Forge AI data management, visualization and analytics, scalability and security. Any final comments you want to mention about Fogi and those various products and how dentists can use them to improve efficiencies operationally.

Yeah. So Fogi has the AI agent component to it, which is, you know, the Forge AI piece. And earlier we talked about the, you know, there's the generative AI and then there's the predictive AI. Well, Fogi has the actual opportunity where it can combine the two in a workflow. So I can take your appointment data and do a forecast with a machine learning model and then actually have generative AI look at that data and make a decision which is very different than what's available out there. So we can combine all of those tools. We also give you, like we talked about already earlier, the ability to have essentially a chatbot that you can ask questions about your data and your practice from the PMS directly and other data sources and. Yeah, and then it all comes to a point and the AI agents being able to access and handle all of these, to be able to actually make decisions and do things. And like I said, we have the turnkey AI agents for the rcm, from insurance verification to claimed processing, et cetera, it all comes together to just make it so that, you know, you can also eliminate tools. Anything from your analytics piece to doing the verification, you can do all within fogi.

Let me ask you a question about Dental Intel. You know, Dental intel, as, as with 80% of dental tech, it sprung out of Utah and is in Utah, the, the mecca of dentists. And I know Dental intel has a verification service. They have. I, I can't remember if it's called the Engagement platform or some other features to it around the rcm. How does FOGI differ from, say, the Dental Intel's process?

Yeah, so we were built as an AI platform first, right. So we started this journey with AI in mind. So everything our product does is designed with the architecture, with the features. Everything is to bring AI and machine learning together and have that at the center of all the things that we're doing. Right. Where these, let's call them legacy companies were built More With a SaaS idea in mind, and they're going to likely have incentives and other things where they're going to be more focused on their traditional SaaS, where fogi is definitely much more focused on AI and has been focused on AI from the beginning. And so you're going to see AI features, you're going to see the value of AI more from fogi versus maybe a more legacy analytics tool.

You know, the analogy this reminds me of a little bit is when Elon Musk came on board Tesla, he didn't, he didn't actually invent the Tesla car. He came on board very early on with some inventors, but he really completely redesigned and revamped the way that the car was built. And they have, in the tech world, they have this thing, you're probably familiar with it, called First Principle Pulse, which is, let's go down, let's. Let's dissect this thing down to its most basic fundamental principles and say, if we didn't have the existing car as we know it, with the combustible engine where it is and the way the doors open and all of that, how would we design it today with the technology and the science and the understanding we have today, if we were to build it from the ground up instead of reshaping it. How do we do it from the ground up a little bit? I see that's the difference here. Where existing companies out there that are involved in the RCM process are going to try to figure out how do we layer in AI to make it better and faster. And that's traditional car companies saying, how do we take our existing car, our existing Accord or Camry, and how do we add in or layer in some battery elements to this? I think what you're saying is we went to sort of this ground up model like Elon did with the car, and therefore it's entirely based on a new way of designing and thinking and building a software tool for automation. That. Okay, that explains it. That's good. Let me ask you this final question and thank you for the explanation on fogi. Really excited to see how it works and to hear some of my clients who hopefully listen to this and get a demo from you and adopt it. Really excited for that. What would you give as advice to dentists who are hesitant or unsure about adopting AI in their practice? Whether it's, man, I just don't want to learn new processes or I don't want to have to terminate or retrain people, or I don't want to have to deal with the risks of potential data management or data exposure issues. What would you say to those dentists?

That's a really good question because it definitely is new and it's, it's. I can see coming from the world of dental, it could be a little unnerving, you know, turning control over to an AI to handle my revenue processes. That the only thing that I could really say is, well, let me give you this example. When I was younger, I actually, my first job out of college is I sold social media advertising. Right. I guess this dates me a little bit because people today would go, well, obviously, who doesn't do social media advertising? But I would actually sell these to car dealerships. And I went, I was in Los Angeles, I drove into this car dealership and I sat down with the guy and I'm like, hey, I'm telling you, social media advertising, it's the future. Everyone's going to be advertising on Facebook, Twitter. Back then, I think there was like Tumblr. I don't know what else there was, right? MySpace maybe, who knows?

And they would look at me and like, you're crazy. We run TV spots.

We are doing. See that billboard up there? I'm like, I get so much business from that billboard. On i5, why would I do social media advertising? And I kind of didn't have an answer. All I could say is like, I promise you, this is the future. If you're going to be left behind if you don't start putting ads on Facebook.

I today I feel like the answer is the same. It's like, look, you don't want to be the car dealership that five years later. If I were to drive down to that same car dealership today and say, are you guys doing social media advertising? Like, are you crazy? Of course we're doing social media advertising. Who doesn't do social media advertising?

I firmly believe in three to five years everyone is going to have some form of agentic AI handling the processes of the dental office. And you want to be the ones that adopted it first, that got all of the early adoption success and not be the one that adopted it later and just got on the bandwagon because it was what of course everyone has to do. Today's the day honestly to jump on the bandwagon because this is when the practices are going to, that get on early are going to win because they're going to see that early adopter benefit and then everyone else is going to have to rush to it because it's just, they're going to have to keep pace with everyone else. So it is scary. You're going to have to jump head first a little bit into the deep end. But I just, I firmly believe from where I'm sitting on this side of the, of the world that like it's going to be a necessity here very quickly.

You know, I would, I would say it is a shift in business management mindset. I don't think it's simply the first ones in are the ones that are going to win. I think it's the first ones in and, and the first ones who make it a continuous part of their operational evolution. So even if you start now and you fast forward three or four years, if you don't continually improve it, evaluate what else is out there, others who may start or adopt later may pass you up. So I think it's more of a mentality that from here on out I am adding AI to the processes within my dental office and I will be regularly evaluating those employees. Just like you said, they're just like employees. I'll be regularly evaluating my AI agents, my AI process and constantly trying to improve that. And I think those dentists that adopted early and then stay ahead by constantly iterating AI in their practice, those are the Ones who I think have a huge opportunity here to be tremendously successful through the adoption of AI.

You have two practices. One has to employ, I don't know, let's say five people at the front. One has to employ two. Right. One can pay less for software products, the other one has to pay more. Like you start to easily go. I think I know which practice is likely saving the most money and which one's probably going to do the most business. So anyways, that's kind of what I'm thinking.

And I would say as long as the AI replacement of the human element in whatever those processes are doesn't cause a reduction in case acceptance, because that's where it could become a problem. All right, I just saved the cost of one employee, but my case acceptance dropped by 30% because the AI just wasn't as effective. Obviously that's a problem. Now the AI can actually help case acceptance go up. And that's when I'll get into some of the clinical side of AI for sort of case presentation and other things like that and follow ups. Then you have the best of both worlds, increasing collections and reducing your overhead. Now maybe we can get back to an average of 40 to 43% profit margin for a general dental practice where it's really slid down to 35% and declining throughout the country because of these compression points in the dental industry. Here at Practice cfo we have formed a tech committee. I find this to be so important that in my own space, I'm a practice owner as well provide financial services. But I have very similar questions and issues that I'm facing, just like dental offices. And so we're forming a tech committee. We're going to meet on a regular basis, we're going to look at our tech stack, we're going to see what AI software out there is introducing itself in the space of CPA and financial planning work. And then we're going to make very deliberate targets and goals to make sure that we're staying on this and implementing it effectively. That's how important I think it is. Now, I don't think a dental office necessarily has to form a tech committee, but I think the doctor and preferably a front office person and maybe a tech, I don't know, an external tech consultant can pop in once or twice a year and sort of do a roundtable think tank on the tech stack and, and, and how to make that better. But it does need to be a deliberate, ongoing, I think, review, just like you do in deliberate, ongoing review of your HR and your team and performance reviews, hiring terminations, things like that. All right, Mitch, this has been fantastic. Very informative. Thank you for coming on the Dental Boardroom podcast. How can people learn more about Fogi and get in contact with you?

Let's go to Fogi IO. Everything you need there. If you want to. If you want to meet me, go ahead and just sign up there and I'll happily find a way to you.

So they go on there, they click a button, and within 30 minutes, all of Fogi is implemented in their dental office through an agentic process. You never even had to get involved.

How did you know? How did you know?

Oh, that's awesome. All right, so go to fo. It's fogi IO. Did I hear that right?

Yep.

Not fogee.com it's fogi IO. And this is Mitch Hershey. Hershey like the chocolate to get in contact with him. Thanks for being on the show, Mitch.

Anytime. Loved it.