XpertDox’s AI-driven medical coding automates claim generation with unmatched accuracy, dramatically accelerating the revenue cycle and eliminating the delays of manual coding. With lightning-fast EHR integration and enterprise-grade security, XpertCoding empowers healthcare organizations to get paid faster, reduce coding burdens, and focus fully on delivering exceptional patient care.
In this episode, we speak with Sameer Ather, CEO and co-founder of XpertDox, about the transformative role of artificial intelligence in medical coding and healthcare efficiency. Sameer shares insights from his journey as a physician turned entrepreneur, highlighting the challenges and solutions in medical coding, revenue cycles, and personalized medicine. The conversation delves into the integration of AI in healthcare, addressing both its potential and limitations, and emphasizes the importance of finding a product-market fit and building trust with clients. The episode concludes with a discussion on the future of AI in healthcare and the importance of balancing innovation with compliance and patient care.
Watch the podcast here:
Listen to audio only here:
Learn more about XpertDox.
Interested in appearing on the SourceForge Podcast? Contact us here.
Show Notes
Takeaways
- AI is revolutionizing the medical coding process to enhance efficiency.
- The transition from handwritten notes to electronic health records has not eliminated errors.
- Identifying pain points in healthcare is crucial for developing effective solutions.
- AI can significantly reduce claim denials and improve revenue cycles.
- Healthcare is one of the last industries to fully embrace innovation.
- Education is key to helping patients understand their health data.
- AI requires a collaborative approach, not just a single solution.
- Building trust with clients is essential for long-term success.
- Over-promising and under-delivering can damage client relationships.
- ROI is critical for the sustainability of healthcare solutions.
Chapters
00:00 – Introduction to AI in Medical Coding
05:06 – The Evolution of Clinical Documentation
09:53 – Sameer’s Journey into AI and Healthcare
14:59 – Identifying Pain Points in Healthcare
20:02 – The Role of AI in Healthcare
24:50 – Future of AI in Clinical Care
29:52 – Navigating Patient Data and AI Integration
35:45 – Challenges in Implementing AI Solutions
41:31 – Understanding Market Dynamics and Client Needs
Transcript
Beau Hamilton (00:00.888)
Hello everyone. And welcome back to the SourceForge Podcast. I’m your host at Beau Hamilton. Today’s episode, we’ll be discussing how artificial intelligence is reforming the medical coding process to improve clinical care for patients and just improve the overall efficiency of our healthcare system, which I think could use all of the health it can get in my opinion. Now I’m joined by Sameer Ather, CEO and Co-Founder of XpertDox, a company that builds tools and helps doctors offices and healthcare systems transcribe messy clinical notes and turn them into accurate medical codes and cleaner claims. And you’ll know that that is no easy feat if you’ve, I don’t know, ever tried to read your doctor’s handwriting. I know that’s kind of a thing in the past now, but it’s an issue nonetheless.
Now, Sameer is a physician-turned-founder whose own clinical and academic background gave him firsthand exposure to the billing and coding inefficiencies inside healthcare with credentials that include an MD and PhD from Baylor College of Medicine, and clinical experience in cardiology. He saw how errors in coding and documentation translate into missed revenue and bogged down the whole healthcare system. It really slows everything down. And I think it’s the classic case of why so many businesses are founded. It’s a problem is identified and a better solution is created and thus implemented. So this is a really exciting topic. I want to get right into it and not waste any more time. Sameer, welcome to the podcast. Glad you could join us.
Sameer Ather (01:45.762)
Thank you, Beau. Thank you for having me over. And again, that was an amazing introduction of the company and my background. So thank you very much for that. And I would just like to clarify, even though things have evolved in health system and health care delivery, when Beau mentioned difficult to decipher handwriting of physicians, we don’t really handwrite anymore. When I came to US and back in 2007 and eight, I was still writing my notes in New York. But these days, almost everybody has moved to electronic health record system. But the interesting thing is that bad handwriting did carry through. And what I mean by that is even today, physicians make all sorts of spelling mistakes. They use their own shorthand and acronyms and all combinations of phrases when writing their notes. And that’s not very easy to decipher, as Beau mentioned. And to be able to structure that information into something intelligible and something that payers and other stakeholders can use and leverage to basically facilitate and make the whole healthcare. We work in the most in the revenue cycle side, but just healthcare delivery and revenue cycle more efficient.
So it is an area where we implemented our platform. I would want to say Beau that, in this case, we were more of an accidental fit, XpertDox and revenue cycle, not an intentional fit. When we started the company, back in the day, I was still training in cardiology. And my biggest issue was how do patients with rare diseases find a doctor that really knows what their condition is and how do they, how do they will find that physician. My background, as you mentioned, Beau, is in healthcare data and artificial intelligence. I’ve been working in the field since 2008 and nine, and I was always very interested in using those tools to really understand this healthcare data, which exists in partitions, it’s blocked off, there is no connectors between them. There’s no data lake that exists out there. How do I leverage that tools that I had to extract valuable information about from this healthcare data?
But our journey has taken us from rare diseases to medical coding, where we did find huge success, as you had mentioned. And I do believe many clinics, hospitals, and providers really struggled to get paid for the services that they provide for their patient. More and more of their time, it now goes into healthcare administration, which should not be the case, right? We want the doctors to be able to see the patients and be done with it. We don’t want them to be dealing with the revenue cycle side and getting denied for whatever they submit.
And when we brought our platform forward to one of the clinics that we knew the founder of, and the ROI or return on investment we saw for the clinic was so immense that we knew we had a really good product market fit and then we decided to scale it up from there on.
Beau Hamilton (08:06.779)
Now, I do want to talk about your kind of journey or how you got started and kind of got into this industry. You’re one of the few founders in AI who also holds an MD and a PhD. I don’t know how you have the mental capacity and the time management skills to get to where you are. I think that’s very impressive, but that’s besides the point. I’m just curious if there was a moment in your clinical or academic career that first exposed you to a lot of these inefficiencies that later inspired XpertDox.
Sameer Ather (08:59.278)
So, no, it’s a great question. I think I was an accidental AI scientist. I came to US definitely to pursue research and education beyond just clinical practice. That was my primary driving force. When I did come and talk to one of my mentors in 2007 or 2008 in Baylor College of Medicine, Dr. Doug Mann, he guided me towards AI. Back in the day, the machines were not as powerful and the softwares and the platforms we had were not as robust as these days. But he said one day, AI is going to be driving clinical decision making. I still remember that’s what he said. A patient will come into the ER, but instead of residents ordering random tests and then doing things, the EHR or the software would guide this new resident because he was concerned that a first year resident who’s just coming out of med school and if the patient is really sick, what’s going to happen? The care may not be as optimal. He’s like, these software will guide the resident in ordering particular tests, guiding them towards management. And he said, I would like you to focus in that area. And that got me really interested and intrigued.
And the Texas Medical Center, it’s an amazing location. There’s 14 hospitals. There’s MD Anderson, there is Baylor College of Medicine, UT Houston Methodist Hospital. I basically did my research in all those hospitals. I was able to build a team of mentors and colleagues across. My first paper on AI, which was on self-organizing maps, came out of Methodist Hospital. All of my advanced statistics courses, predominantly in the field of stochastic gradient boosting, happened at UT School of Public Health, one of the leading institutions in the field of public health and data science. And a lot of my clinic, not a lot, all my clinical work was at Baylor College of Medicine. My clinical informatics class was actually at Texas Children’s Hospital.
So I would say I leveraged whatever I got, whatever environment I had, I think I leveraged it to the hilt and was able to get the clinical training and the research experience that enabled me to combine both of those to understand the problems in clinical care, delivery, access, and processes, and identify how, with my background in AI, I could potentially solve some of those problems. So I think I’m lucky to have that cross experience between the two of them.
Beau Hamilton (11:40.763)
Well, that’s neat that you have all those resources right there and that Texas hospital area. Like you had all those 14 different kind of hospital branches on that one area, geographical area.
Now what was the time period when you were kind of first getting directed and told about this kind of AI future that the healthcare industry is working towards? Because I know XpertDox has been around for about a decade, right, in some capacity.
Sameer Ather (12:34.094)
Yes. We started out Beau in 2015, but as I mentioned, we were predominantly on the rare disease space, space and clinical trial. We still have Stanford Cancer Center as one of our partners where our platform is enabling patients with cancer, connect with the appropriate clinical trials going on at Stanford Cancer Center. We also have Lifespan Health, which is part of Brown University. We previously had University of Alabama in Birmingham. While we were there, we honed a lot of our skills in data parsing, natural language processing, standardizing diseases into nomenclature, which back then was ICD-9, now it’s ICD-10, but also other terminologies, both like mesh terms are also used to describe diseases and SNOMED terms. So we developed a lot of skill set and processes on our previous sort of products that we had.
From 2008 and 2015, I was either doing research or clinical training before then in cardiology. And I think I published more than 50 research papers on the topic of personalized medicine and statistical genetics. And the goal at the time of all my research was how to leverage AI to identify which treatments and management would work for a single patient and would not work. So that’s what the definition of personalized medicine is. Personalized medicine, in traditional health care, we bucket a person based on a few things. So we would say, you have diabetes, you need to take these three medicines, right? That’s what we do. A personalized medicine says, you are different from all of the diabetes patients. You are you. You are defined by your phenotype. You are defined by your genotype. You are defined by your metabolome. You are defined by your environment and everything else. You are you. So every treatment that you get should be personalized to yourself. As the way we are set up, it is difficult to take in those millions of data points to predict what will happen next. So that is the goal of one size medicine and statistical genetics. And that’s what my research focus was.
Beau Hamilton (15:00.881)
So that’s where, so you’ve really focused on the personalization aspect and then, but you also, you focused in on obviously the medical coding and maybe even kind of the revenue cycle topics in this healthcare space, right? Which I think aren’t necessarily the most like glamorous areas in the healthcare system, but are super important as you have alluded to and noticed firsthand.
Maybe, and you did touch on this, but I’m curious, like, what made you focus on this specific kind of pain point rather than something more traditional or technical like diagnostics or other workflow automations?
Sameer Ather (15:40.846)
Now Beau, that’s a great question. I would also say if you have listeners who are looking to start a company, it’s very important to find a product market fit. And many founders are actually blinded by the needs of the market. They are usually driven by their own belief systems and the expertise that they have. So they want to build a solution that fits the market rather than identifying what the market needs and building a solution for that.
When we in 2021 and 2022, around that, we noticed that there was a huge surge in some of these urgent cares. There a lot of patients who are coming in with COVID or COVID testing. And these clinics were unable to submit their claims to the payers and they were late by months in submission. And what it means is if you don’t submit your claims, you will never get paid. If you don’t get paid, the clinics are in trouble. We had an urgent care chain that was facing this problem and they asked us to build something for them. And we debated, but we did build it for them eventually. Once implemented, we figured out that revenue cycle is a big problem. It’s cost heavy. So not only do the providers have to pay for the space and pay themselves and pay the nurses and pay everybody else. But now they have to spend time for a team to code their claims or build their claims. It’s a huge revenue cycle. The revenue cycle is costly. Not only it is costly, but many claims do get denied. If I remember reading all these headline numbers, 30 to 40% of the claims get denied in the first pass. Half of them are because of coding and billing errors. That is a travesty. That should not be happening. And now these providers have to go back and fix those and submit again. And if the coding is not accurate, then the wrong patient is being hit by the wrong copay. You don’t want that. That’s a patient satisfaction is also an issue.
So a lot of problems that we saw once we implemented our minimum viable product in 2022. there were so many pain points, by the way, that we have been growing at a tremendous pace and have been creating a lot of value for our clients. We recently had another press release, I think two days back with CHP box shares. They saw a huge jump in the revenue. Denials at our partners run to close to zero. It’s not a 30% reduction. We talked about 15%, 20% claims being denied due to coding errors. Our denials, when they’re quoted by our system, run at 0%. That’s less than 1%. So, we have been able to, we have been lucky, have been at good, great partners, not good, but great partners. And we have been able to generate tremendous value for our partners over the last three years.
Beau Hamilton (18:49.287)
Well, that’s great to hear and it just, it makes me kind of sympathize with the problem you’re trying to solve because you impact so many different kind of hurdles. There’s so many hurdles you have to navigate. And then also you have to obviously appease kind of like the clinical staff insurers, but also patients who are ultimately affected by this whole kind of nightmare situation of getting their procedures delayed or denied because of coding issues. So there’s a lot of, I think it’s great that you’re trying to tackle this issue though. I think it’s gonna have, obviously it has a real world impact with how it affects patients directly or indirectly.
Now you’ve spoken quite a bit about AI, I know that like we’ve in the past, we’ve talked about the limits of AI and healthcare. We have this like new exciting technology come along. It’s easy for us to kind of get ahead of ourselves with all those sort of promises expected to provide society and impact all these different, you know, healthcare industries. It’s, the main reason I think why there’s so much chatter about this like AI bubble because so many billions of dollars are being thrown at this technology. But from your point of view, as a physician, where does AI truly add value? And then on the flip side, where does it maybe fall short?
Sameer Ather (20:18.444)
A difficult question. And I would not like to I can’t comment on the business side of the whole valuation of all the companies and stock markets and everything else. I would say when I read the literature and historical records and if you look at the dot com fase, right? I was in high school when this was going on. We are living in a distillation today, right? But we still had a dot com crash. So at some point, there will be an over promise. And maybe there will be a connection at some point. Again, not predicting anything. But does, will the AI transform how we live lives? I think it will. It would. And will. To what extent and what that timeline would be is, I think, up in the air, it might take a long time. We are thinking like everything is going to change next year. I think even the dot com, if you think about this, right? We talked about super builds in 2015, 2016, I was still using them in my clinic. Whereas a dot com was starting in 1990. So it took decades for everything to become digital. Maybe I will take less time. Maybe it will be 10 years.
Next question you had was what’s really the scope, right? Do we will we have humanoid robots walking around doing everything. I have not seen an evidence for that yet. Would it happen? We would all be actually, we’ll all see together, right? I think the next part of it was, if your question was pertaining to how do I see AI specifically in healthcare? I think there is in healthcare, it’s one of the last places to be touched by innovation in general, apart from biotech and the drugs that we’re developing, which is an amazing work a lot of pharmaceutical companies are doing.
But if you look at healthcare in general, healthcare is always focused on safety. It’s always there’s issues with compliance. The patient health information, PHI data, HIPAA compliance, a lot of the regulations happening that AI is, healthcare would be affected last because of the safety issues. But I do think because of that, I also think that healthcare has the most to gain because there are still so many processes that have not been touched yet by even the technology we have today. But we don’t have to wait for the humanoid robots yet. We can take it from intake. I’m quite sure you get frustrated by finding a doctor. You are frustrated by, say, you have an appointment time, but you are not able to. When you go there, you still have to wait. Or maybe who is within your coverage area? Which procedures are there?
Let’s talk about transparency on cost, right? Such a big emphasis for last 15 years that we’ve been talking about. So when I envision AI, I think AI will touch everything from patient intake to clinical care delivery, which my mentor did talk about, which is it will guide clinical care. It will guide which tests to be done, what treatment is applicable, up to the point of what we are touching right now, which is the back office, which is once the clinical care is done, you have to still complete all the processes for the cycle to be complete, it will touch almost every aspect of healthcare. First would be more with a human in the loop model, where it would actually enable humans to do things better. But then in piecemeal, will be able to automate things without supervision, but in piecemeal. Full automation, I don’t foresee, at least I don’t foresee. Maybe in five years I will be able to foresee and there.
Beau Hamilton (24:30.631)
Sure, Yeah, no, it’s fascinating. I always love hearing founders insight view. And I think it makes sense when there’s, you said this is kind of the area where there’s maybe the most to gain from AI improvements. But you also mentioned like the regulations and compliance issues that you do have to kind of navigate, which are there for a very good reason, but they do kind of slow the adoption rate down a little bit because you want to make sure that when you adopt this technology, privacy is of most important, security is there. And I think a good visual example is if anyone’s gone to the doctor’s office in last six months to last decade, really, you see this move to the cloud-based, like MyCharts. And a lot of this technology is still, I mean, a lot of my doctor’s clinics are have adopted the cloud-based platforms relatively recently.
But it’s such a big transformative adoption of technology because you have this central location where you can keep record of appointments, your medications, your notes, you can talk to doctors directly. I think that is a huge, that was a pivotal technological breakthrough, just having that platform to go to and then being able to have other clinics tie into it and interact with it. So I think having that platform is good. And then I just imagine the various AI automations that kind of get baked into a platform like that, both from the patient’s perspective, as well as the clinic’s perspective in the software they’re using to manage everything.
Sameer Ather (26:13.893)
I think so. Well, what I’ve seen is that there’s been a push, of course, from a patient side, and then it has been adopted by clinics and providers is to be able to provide all of their documents pertaining to the clinical care of a patient back to them. And that is I would say I’ve seen that shift in the last four or five years. It does come with a little bit of drawbacks.
I was recently pulled aside by our previous nanny of my kids. She said, I think I’m gonna die. And I was like, what happened? It’s like, I have a kidney problem and it’s not gonna work out. I was like, what’s on the records? What’s going on? Her creatinine was barely, like it was 1.0. The cutoff was 0.9. It’s completely fine. She’s 65 years old and she’s fine. Think about this, till she has the next appointment, she was scared that she has a kidney problem and that’s it. And so as I started in the beginning, there are always trade-offs. I still do think that trade here, the benefits far outweigh the risks. I think we do want to make sure that we educate the patients not to over-interpret the data that they get from these MyCharts and to wait to speak with their providers to really get the insight on what’s going on.
Beau Hamilton (27:50.203)
No, absolutely. It goes like, if I have a problem, a medical issue, or just a life question, I mean, nowadays I’ll just, I mean, back in the day, I’d go to Google and see what kind of find some forums, some Reddit posts. But now you have these chatbots, you can plug in your question and they’ll try and they’ll do their best to give you some recommendations. But a lot of times it does kind of give you sort of the worst case scenario. I bet you’ve had problems or just struggles dealing with patients who have done the same thing come talk to you and be like, am I, you know, am I, do I have like this, this dire, terminal kind of, outcome of diagnosis or? What’s your experience been like from a doctor’s perspective and seeing all these, these patients and relatives, friends use these, these chatbots for medical advice.
Sameer Ather (28:43.544)
I’ve been very lucky Beau. I see patients at the Birmingham VA Medical Center and there’s our veterans. I’ve been in the VA now for 15 years. I love my patient profile. They’re always thankful. They listen to what I say. They may or may not do what I say, but they at least listen. I just have a very good patient population.
And I think if you are somewhere out there doing more of a boutique or practice, I suppose you get hit with those questions a lot more. I’ve been lucky. But I also think that it never hurts. A question never hurts. A question can only help something, right? It could potentially irritate a provider. But I always think not asking a question can be a risk. Asking a question is never a risk.
Beau Hamilton (29:36.059)
And then also just, you’re a patient looking at the information, take it with a grain of greatest sand, greatest salt. Don’t stop seeking other opinions. Don’t cancel your doctor’s appointment and visit just because you see something that’s generated from one of these chatbots.
Now, I imagine as far as XpertDox go and some of the features you offer in your platform, I’m sure there’s like a laundry list of different functionality you’re considering adding to your platform, given just kind of the scope of what your software has the potential to do and what it currently does. How do you prioritize which features to add? And then how do you build an AI platform that has to integrate with so many fragmented systems out there? Because I mean, some of this we’ve discussed, you have the MyChart, you have electronic health records, you have the compliance risk management issues to consider. There’s the just inconsistent quality of documentation, various coding standards. There’s a lot to work with and integrate with. So how do you go about trying to work with all these systems?
Sameer Ather (30:44.408)
I still think even after the Affordable Care Act and everything that we have done over the last 15 years to build interoperability, I don’t think it has materialized yet. A health system still remains, healthcare still remains very fragmented. And so we use a variety of tools to our automation team uses to connect with the EHRs and the PM systems and other databases, including robotic process automation, HL7, APIs.
And so we try to connect with the EHR, whatever’s easiest for the client. We don’t dictate it, whatever is the easiest path to go live. That’s what it is. Because many clinics and providers are not very IT savvy, and they don’t have the IT resources. So we do not mandate that this is the way we do things.
So I would again break it down into two parts. One is we build what clients need in us. We don’t really try to dictate what we want to build. I’ve made a mistake in the past. That’s why we now focus on clients. Clients tell us what they want, what they need. And if we continue to do a good job, making sure that we make their life easier, we generate revenue for them and we save costs, I think we’re in good shape.
When it comes to what platforms and aspects of medical coding are out there Beau, well, it’s a very interesting field. It’s broken down into three main components. When you go see a doctor and a doctor provides a small procedure, let’s say puts a splint on your hand, you broke your arm. And they ask the insurance company to pay for something. They may say, I saw them, they read an x-ray, pay me for an x-ray, pay me for the splint, pay me for reduction or something, right? This is called fee for service coding, is whatever care you receive from the provider, providers asking the insurance company to pay them for all the care that that person or clinic has provided to you. This is called fee for service. What’s a downside? As I always say that here we are going to use this buzzword, right? Trade-offs. There’s always a pros and cons. Healthcare costs in US have continued to skyrocket. We all know that, right? So the payers and the regular, everybody has thought like, this is not working out. There’s too much care being provided when it’s not needed. What do we do next?
So they came up with a system called risk scores or value-based care. And they said, fine, we will pay you a fixed amount of money per patient for the whole year. And we will determine how complex the patient is. So our diabetes would make it more complex or hypertension or obesity or something else, a combination. So they created this risk profile. And they started giving these payments out to the clinics for the whole year to take care of this patient.
They saw another issue, which is the amount of care being provided actually went down even more than it should have. So how do we, when then the third part as piece of it came through, which has now been plugged in to both fee for service as well as value based care coding, which is quality metrics. So fine. Even though, even if we pay you fee for service or we pay you by managing the patient, here are the targets. If you have a patient with diabetes, we want to see a hemoglobin A1c below seven. If you have a patient population with a lot of hypertension patients, we want to see a blood pressure below 140 or below 90 diastolic. If we have patients with this, that, whatever it is, we don’t want to see antibiotics being prescribed, when you have a flu. But we do want antibiotics to be prescribed when you have a strep throat. So there’s lots of quality metric coding that is also in place that determines the amount being paid back to the clinics and for that’s to incentivize them to provide really good quality of care.
I don’t think at this time we have figured out what percentage of each is the appropriate mix. And it could be that, that mix would be different for different patient population. For nephrology or dialysis patients, maybe the value-based care is really good, but for urgent care, it could be something different. Coming back, and I don’t want to continue on this, medical coding does take on multiple aspects. There’s more complexity to it. And at this point, we have built out the platform to take care of all those aspects for our partners.
Beau Hamilton (35:53.649)
Well, that is really interesting to learn and hear about the behind the scenes kind of coding that is involved. And also it does go back to the complexity and then just how the personalization of working with clinics and each patient has a series of issues and treatment requirements. And again, it complicates the whole sort of solution you’re trying to come up with, but I’m glad that you’re obviously, you know, hearing this feedback firsthand and working on a solution case by case, it sounds like.
And you mentioned the sort of maybe it troubles when, working with a clinic to adopt your platform, I’m curious, when, when you are maybe approaching a clinic or viceversa, trying to get, a new kind of, customer client as adopted, set up with your platform. What, what’s some of the pushback you receive? Like I imagine it’s less about costs and more about just burden and like the learning curve associated with it, right? What are some of those pushbacks you’re getting?
Sameer Ather (36:59.406)
It has evolved Beau. 2022, if we went to a hospital or a clinic or a practice, there would be the first question would be, is it safe? Is it secure? And what if it makes a mistake? What’s what’s going to happen then? Today, it’s very different. It is very interesting. Most of our clients, almost all of our clients approach us, we don’t approach them. I think that one big thing that has happened with Cloud and Anthropic and ChatGPT and Gemini and all these platforms, I think the comfort level has increased quite a bit. And I believe it’s being driven by their boards and the CEOs under pressure to implement AI solutions to seek efficiency, cost reduction and improving revenue. I think their careers depend on it.
So they are seeking out solutions within their organization, how to leverage AI agents, platforms and softwares to find those efficiency gains. So, well, it’s completely changed over the last three years. We were out there trying to convince people three years back, and now people approach us that we want to implement your solution. We just have to figure out which one to implement, but we will implement something or somewhere.
Beau Hamilton (38:28.551)
Now, when it comes to the actual automation components and maybe after you factor in which implementation to go with, when you talk to these clinics, what’s one of the biggest misconceptions they have about your solution? And then maybe broadly, the AI-driven automation abilities that are coming down the pipeline? What are some of their concerns that is kind of maybe misguided?
Sameer Ather (39:08.526)
So the way I would describe it is that AI traditionally has been a marketing tool. When we use the word intelligence, we’re trying to put a human aspect to a software or a code. When we say machine learning, learning is a very human or an animal or living organism behavior, but we use it for softwares.
And I find it a little bit sort of disconcerting having been in this field for last, what, 16, 17 years now. I think when people approach AI, they really think that there is some sort of humanoid thing sitting in the background doing all the work, which is very intelligent, very smart, a mini Einstein doing everything. But then we have to inform them that to build a platform like this requires a team effort, not just team, teams effort. We have an automation team, have a data analytics team, we have an informatics team, we have an audit team. Each team member has a different role in there. To be able to successfully code the claims, it has to go through multiple processes, multiple checks and balances. We also do not use any single AI tool. We use multiple, that’s why we call it a hybrid AI. I’m a big fan of gradient boosting, I use that a lot.
We use transformer neural networks a lot. We use rules-based systems as the final checkpoint, and we have an audit human in the loop there too. So we take accuracy very seriously. But coming back, I would say when clients or partners or leads talk to us, they believe that they have a solution which they will click on and somehow this, again, as I said, a mini Einstein will start come and do stuff and then we have to tell them that now it’s a process. We have to go through an onboarding and the automation team has to understand how to get the data. Then we have to build processes. It’s a teamwork.
Beau Hamilton (41:19.053)
Yeah, no, that makes sense. They’re almost kind of looking too far maybe into the future and also looking through the lens of kind of the doom and gloom approach where everything is just completely automated and there’s no human in the loop. But yeah, you got to go through the right processes. You got to go to the right channels onboarding. Make sure that you actually understand their problem and kind of educate them.
I think a lot of it just comes down to educating the clients with what this tech can do and how it’s the future and how it has this end. The end result is helping you become more efficient and helping in the long run. But there’s just like with anything, you got to kind of go through the right steps first before you get that goal, that end goal. But I think that makes you kind of uniquely positioned because of your experience understanding firsthand kind of some of the issues that you’re, you’re, that you’re seeing clinics struggle with. But then also the taking the AI experience you have with neural networks and modeling.
Really, really interesting. I have a couple more questions. I want to know like what you’ve learned about your market or your customers that have helped sort of change your approach, or if there’s something maybe in particular that your competitors are getting wrong, that you’re determined to do differently. Anything that comes to mind?
Sameer Ather (42:47.842)
I won’t be able to speak completely about the competition because again, I am not in there. I can only relate to things that I’ve heard my clients talk about our competition. One is that I believe two things, over promise and under deliver is a big problem.
If I have heard clients switch to us or when they talk about other implementations that they have gone through, that the sales team somehow over promises things and then fails to deliver. And the second thing I would say that why we are, and we always sort of under promise and over deliver. And if there’s anything in the, that we do, we actually put it on our service agreement as a service level warranty that you will see a denial due to coding error as less than 2%. You will see all claims being submitted within 24 hours, or you will see an automation rate above 90%. That’s our service level agreement. So if we are going to promise, we make sure that we put it on paper for them.
The second aspect of it is I think we are very nimble. We try to understand the clinics own issues that they have. And we try to adopt the platform to make sure that they are seeing an ROI from the implementation because we do believe that without that, they won’t be long-term partners to us. And I believe even at SourceForge, right? you have already seen some of the testimonials that our clients have put about us. We’re very proud of the work we have done with them.
Beau Hamilton (44:49.955)
Absolutely. Yeah, that’s a great answer. think the over-promising, under-delivering is something that anybody can relate to. Every industry has these kind of, I don’t know, like sales pitches and clients and companies they’ve worked with that they do over-promise and they’re not, and at the end of the day, they’re not satisfied with the end results. And it’s especially true in the tech sector because we see this shiny new kind of future to look forward to, but we realize like, there’s a lot more steps involved to get there than one might realize. And even from a company standpoint, from the tools that they have under their belt, can, the marketing gets kind of gets ahead of themselves, I think.
But great, great way to explain it. I think at the end of the day, it does just come down to building, building trust with the relationships you have, with the clinics, with businesses in this space, and then just letting the results speak for themselves. And then I think on top of that, underscoring the fact that, you know, this is a win-win for clinics. This is a win-win for patients once this is properly implemented. If patients can get better treatment and be seen more frequently, I think, you know, especially at a lower cost, I think that should be the main focus.
Sameer Ather (46:06.512)
I completely agree. If there is no ROI for clinics and providers and hospitals, no solution will last for a very long time. I think that’s true for not just for, it’s true for in general, right? You have to focus on the clients. You have to make sure that they are seeing the efficiency gains that you have promised. And if not, then it’s not going to work.
Beau Hamilton (46:28.837)
Absolutely. Well, Sameer, thank you for all the insights you’ve shared. For listeners curious to learn more about XpertDox or get in contact with you and your team, where should they go?
Sameer Ather (46:41.111)
Well, we have our website that details all the information about us, our partners, how we work, how we integrate the technology that we use. Actually, we publish all of the live numbers on our landing page across all our clients every day. So if somebody was interested, they could take a look at our website. It has all the information, but if there’s any information that is missing, you can always reach out to us at info@xpertdox.com.
And really appreciate Beau for spending this time and chatting about healthcare in general, the intersection of AI and healthcare. So thank you very much.
Beau Hamilton (47:17.635)
Absolutely. I love it. Well, thank you so much for everything. Sameer Ather, CEO and Co-Founder of XpertDox. I’m really excited with how this turned out and I hope we can have you back one of these days, follow up with some of the things you’ve discussed.
Sameer Ather (47:31.447)
I would love to Beau anytime, Thank you very much. Thank you for having me.
Beau Hamilton (47:37.177)
Thank you all for listening to the SourceForge Podcast. I’m your host, Beau Hamilton. Make sure to subscribe to stay up to date with all of our upcoming B2B software related podcasts. I will talk to you the next one.