Episode #250: Sergei Azernikov – How Technology Is Transforming Modern Dentistry

Dr. Eric Block: Welcome to the Stress-Free Dentist Podcast. I’m Dr. Eric Block. If you’ve ever felt overwhelmed, burned out, or like your practice is running you instead of the other way around, you’re in the right place. Each week, I sit down with leaders in dentistry to share the lessons they’ve learned from their successes, failures, strengths, and setbacks. The goal is simple: to help you build a more efficient, productive, and less stressful career, and a better life outside of it. For more resources, books, and upcoming events, visit thestressfreedentist.com. Now, let’s get into today’s episode. Hey, everyone, welcome back to another episode. Today I am joined by Sergei Azernikov, who is the SVP — Senior Vice President — of Software Development at Glidewell Laboratories. Sergei, thank you so much for joining us.

Sergei Azernikov: It’s great to be here, Eric. Thank you.

Dr. Eric Block: Actually, Sergei, we just saw each other not too long ago — a couple weeks ago. You were nice enough to invite me on your show, the Glidewell TV show, and I had a great time. You’re a great interviewer, and you’ve got a great story. One of the nicest and smartest people I’ve met, but also extremely humble, and you really welcomed me to Glidewell in your studio with open arms. So let’s first just hear your story — how did you go from where you were to working for a dental lab?

Sergei Azernikov: It wasn’t a straight path. Even in school — my master’s and PhD program — I had some exposure to the dental space, because I was in computer vision and computational geometry. This was the early 2000s, when technologists were just starting to get into digital technology, but it was really early. I was in Israel at the Technion, and an orthodontist came to our lab. He was forward-looking enough to understand the potential, and he came to see what was available in scanning and 3D printing. We had some very early scanners — nothing close to what we have today. It wasn’t possible to scan a patient back then, just casts, like stone models. But he already understood this could be taken to the next level. He even asked me to present at a seminar he was giving at the Hebrew University in Jerusalem, to explain to the students what was available and what was coming. That was more than 25 years ago. He told me that the company that came up with an intraoral scanner that actually worked would be a $1 million company. Now, 25 years later, we have billion-dollar companies that do exactly that. But that was really the start — the first exposure. From then, I didn’t think I’d actually end up in dentistry, but I knew there was an opportunity. I moved on. I worked on some industrial applications at the University of Michigan as a postdoc, then took my first job out of academia with Siemens Corporate Research, working on hearing aids — patient-specific, but a very different vertical. From there I moved to other patient-specific applications, like orthopedics. And, unsurprisingly, the workflow in all those verticals is actually very similar.

Sergei Azernikov: It’s patient-specific — some sort of modality acquisition of the patient’s anatomy, design of the device, more or less complex depending on the case, and then some sort of manufacturing or rapid prototyping, with 3D printing or milling. That workflow is pretty standard across all these verticals. And that’s how I got to dental. I saw that Glidewell was building a new team to work on dental design and design automation — which is what I’d been doing before. So I realized this was another vertical to get into early, using similar tools, and innovate there. Interestingly, 3Shape — everybody knows it today as a dental company, but I knew it as a hearing aid company from my Siemens days. Coming to Glidewell, I saw that Glidewell was actually using 3Shape, and 3Shape literally learned how to make teeth from Glidewell and built their product initially for Glidewell. But Glidewell quickly realized they needed their own software to control their destiny, to expand quickly, and to do things that fit their scale and pace. So when I came on board, it was as a developer — a software and algorithm developer, which was my background. It was really the beginning. I was the second person on the team. Today we have around 200. So we’ve grown quite a bit. It’s been over a decade — about 14 years now.

Sergei Azernikov: Later on, we saw great potential in machine learning and AI, when it really burst into the common space. Around 2015, it became clear the tools coming up could be very useful in many applications. We realized the first potential in dental design — automating the design of crowns and bridges. And today, we have 100% of our designs done using generative AI. Everybody’s talking about this now, but back in 2015 and 2016 it wasn’t obvious. Those advancements actually made waves outside of dentistry as well. We presented at conferences like NVIDIA GTC. Today everybody knows NVIDIA — it’s a $5 trillion company. Back then it was a very small event, maybe 5,000 people, and the company was known as a gaming company. Today it’s all about AI, but in those early days it was still mostly gaming and a little bit of AI. So we were very early adopters of this technology, as well as cloud. AWS came about at almost the same time, and cloud allowed us to accumulate a lot of data and make it available for machine learning. It turned out dentistry is a great use case, because you have lots of data that’s kind of similar but different. Even the same person doesn’t have the same teeth — all teeth are a little different, but close enough to learn patterns. That’s exactly what machine learning is about, so it was extremely effective. That’s the high-level journey. Today I’m running the whole development — it’s not just machine learning, that’s just part of it. It’s the systems that support our manufacturing, our corporate side, the designers, and the chairside that you’re one of the more prominent users of, which is powered by the same software we’ve been working on.

Dr. Eric Block: Now, when you were working for Siemens, did you actually scan the ears for hearing aids?

Sergei Azernikov: That’s a great question. Traditionally there was an impression. Unlike teeth, which are positive and where the impression is negative, with ears it’s the opposite — the ear is negative and the impression is positive. So it’s easier to scan using optical scanning. The common practice was to make an impression and scan it with an optical scanner. And, again, the first 3Shape scanner was made for scanning ear models. Later on, there were attempts to come up with an intra-ear scanner. That’s harder, because the ear canal is concave and very complex — some people have a more curved canal, some less, but in general it’s curved. It’s also very important not to hurt the membrane, so there are lots of constraints. It’s a fairly hostile environment, with cerumen and hair, so it’s not easy to scan. There are some scanners on the market, but I’m not that close to it right now. As far as I know, the common practice is still to make impressions and scan them optically, because it’s easy — whereas scanning a dental impression optically is hard, or sometimes impossible. That’s why we developed our own CT scanner for that. With ears it’s actually easier, so that’s the standard of care.

Dr. Eric Block: Interesting. You said you’d been working with AI since about 2015. We’re in 2026 — that’s 11 years. How much have you seen it change over the last 11 years?

Sergei Azernikov: The pace is actually accelerating. It’s changed a lot, but the pace is becoming so quick that it’s now changing almost weekly, or maybe daily. What we were using back in 2015 were very early generative models — very far from the size of the models people train today. Today the focus is on models that generate text — large language models — and they can be applied to such a variety of applications, because a lot of our routine tasks are related to language. In the beginning, we were looking at models from the computer vision field — they generated images or 3D models as we needed. There were different types and sizes, very small relative to what we have today. Back then, hosting and using those models wasn’t easy. Today you don’t even need to train your own — you can use them off the shelf. That wasn’t the case then. We worked hard on collecting data, training our own models, and even hosting them somewhere easy to access. Today we have AWS, Bedrock, and all the systems that allow very easy management of ML models. That didn’t exist back then, so we had to build that infrastructure ourselves. Things changed dramatically. Today the entrance barrier and the costs are different — you pay for tokens using somebody else’s models, rather than hosting or training your own. Very few people train their own today; most just use what’s available. So the use cases changed, the availability changed, but one thing hasn’t: it’s progressing super quickly and enabling more and more innovative applications that open opportunities that weren’t available even a couple months ago.

Dr. Eric Block: That’s amazing — how quickly it’s changing. I see it on my fast mill; the crown proposals I get are just getting better and better. How are you using AI with these improvements? I imagine you can code and make software developments much quicker now.

Sergei Azernikov: Right. There are two vectors of improvement here. One is that the way we train and generate those models is improving — and you see those results. The other is coding — the models and systems others are putting out, the LLMs that came into the spotlight. Systems like that let you build and code much quicker. So it’s essentially a self-accelerating wheel: you develop faster, you get more adoption, you get more data, and then you get better models and better utilization. That’s why the pace of change is so high today. Initially, when we wanted to build the first version of the generative AI that created crowns, we spent a very significant amount of time — a couple of years. We started with a collaboration, which is interesting to mention — with UC Berkeley — and that took about a year just for the proof of concept, to see that it could actually work. Then another year to productize it and make it fully utilized in production. So from 2015 to 2017 — that was the timeframe. Then it took a couple of months to select the right data for training, because you don’t want to feed all your designs into it; you want the best designs, done by your best technicians. We were lucky that our owner, Mr. Glidewell, spent literally hours with us every day reviewing crowns and quality and advising us on what to include and exclude. That all took a very long time. Today, updating a model might take us one week, instead of months of work. That’s the scale of change — and it can potentially get even faster.

Dr. Eric Block: Wow. I imagine everything’s just going to keep getting better and quicker — the time it takes to mill a crown, design a crown, or trace the margin. I’ve noticed it since I got the fast scan and fast mill: it’s gotten better, quicker, the machine’s gotten quieter — these little improvements that keep happening with updates. How do you envision the future here? Are we just going to keep seeing updates and improvements like this?

Sergei Azernikov: From the start, we had a vision of a system where the doctor just needs to scan the patient. Once the scan is done and it’s high quality — good enough — the system can take it from there and design the restoration. You just look at it, maybe make minor changes, but ideally no changes are needed. Then it goes directly to the mill and into the patient’s mouth, because it’s already pre-centered. That was the vision from the start, and we’re getting closer and closer to it. We make the mills faster, we improve proposals, we improve margin detection, and working with our partners, they improve scan quality. If you look at the quality from, say, the i700 to the i900, there’s a real leap — the input becomes better, the proposals become better. I believe we’re very close to a point where it’s really a one-button click, which tremendously reduces the entrance barrier for dentists to start chairside milling. I don’t need to convince you — you’re one of the people who really believed in it early. But a lot of doctors still don’t get into it because they hesitate. They either don’t want to spend the time to learn it, or they don’t think it makes sense for them. Once it’s highly or fully automated, though, I think it’ll be a no-brainer. That’s the vision, and we’re constantly pushing the limits of it. We’re actively working now on integrations with practice management software to streamline the process even more. Today, you create your treatment plan in your practice management software, but then when you go to your scanner, you have to retype all the information, which takes time. We want to eliminate that and close that gap. On the feedback loop, we also want to automatically update the status of our lab cases and verify the quality of the crowns we send. If you take a postoperative scan and there’s a problem — for example, an open margin — we can feed it back into the algorithm and improve quality next time. So we’re working on all of this actively. The overall vision is to make it as streamlined as possible, so it becomes just a one-button click.

Dr. Eric Block: It’s amazing. I love it. I love same-day crowns, I love milling, I love scanning, and I can never go back to traditional impressions. It’s made me enjoy dentistry more, and it’s made me a better dentist. When I scan and blow up that tooth and evaluate my prep, I trim out the areas I need to fix, go back, re-prep, re-scan, and do that several times. And guess who else senses this? The patient. The patient senses you’re using the best technology out there — they’re absolutely wowed by it, and those are the ones who tell their friends, family, and colleagues to come see Dr. Block and become a patient. Some dentists are very hesitant to change, because in clinical dentistry we don’t want mistakes or redos — we do things we’re comfortable with. But if you’re not moving forward, you’re going to get lost. If you’re just on that hamster wheel doing the same thing over and over, you’re going to fall behind. Maybe that type of dentist is doing awesome dentistry and loves what they do — that’s great. But your job is the opposite: at Glidewell, your job is to constantly change, constantly improve, and constantly get better. You’re almost the opposite of a dentist who doesn’t want to change.

Sergei Azernikov: That’s right. And it’s not just me — it’s been Glidewell’s motto: constant change. There’s a book Jim Glidewell wrote by that name, Constant Change. What differentiates Glidewell, and what I think primarily allowed it to become what it is, is that constant look for the next thing. We’re constantly looking for that next leap — not just iterative improvements, but the next browser, the next chairside system, the next CAD/CAM. I believe that’s what really moves the needle forward. My job in that sense is exciting, but also difficult, because a lot of times the things we do take literally decades until they’re fully adopted. I’ve seen that internally and externally, because we also develop tools for our own internal use. Change is difficult for everybody. But people who are excited about it benefit the most — early adopters benefit the most, as always. They take risks, and they’re rewarded most of the time.

Dr. Eric Block: Now, what’s a day in the life of Sergei? You probably wake up at five, go surfing, go for a swim — but when you get to the office, you have a team. How does it work? Do you mostly listen to customer feedback on something you want to improve, like tracing the margin or the design of the crown? Or do you have one team on that, one team on crown proposals? How does it work?

Sergei Azernikov: Typically, my day starts when I’m driving to the office. I have calls with the remote teams, because we have teams in different geographies, so I start early to overlap in time. At seven or eight in the morning, while I’m still in the car, I’m talking to those teams. We have quick standups just to understand if there are any blockers we need to resolve immediately — that happens pretty much every day. Then it depends on the day. There are lots of days with immediate things to resolve, because I’m now responsible not only for the CAD/CAM side but also for some enterprise systems. Speaking of change, we’re moving from paper invoices to digital invoices as we speak, and there are people who don’t like it. So we need to deal with that — how do we improve our systems while respecting the needs of people who still want their paper invoices? Those are typical challenges. I did it before we jumped on this call, and I’ll probably deal with it after. These things happen all the time in the background. At the same time, we have a true R&D effort going. We have teams for chairside, teams developing tools for our lab use, and dedicated people doing machine learning — I work with them quite closely. Then there’s a whole team of QA personnel trying to verify everything we put out there. We also have a big push now on user experience — improving the systems we give our customers, like our portal and mobile app, all of which need to work together with our chairside system. And, like I mentioned, integrations with other systems are a big focus of mine right now. So all of this is moving at the same time. I feel a little bit like I’m in the circus — juggling five or six different tasks at once. My calendar is full of meetings, and the time I can really spend thinking and doing deep work is typically after everybody’s gone, usually after five o’clock. I lock my door and just think about what happened during the day and plan for the next. And I try to squeeze in a workout, which I think is very important. As I age, that’s becoming more and more critical to be able to do all these things and stay focused.

Dr. Eric Block: Your day sounds very similar to mine. Instead of calls in the car, I have a morning huddle where we review the day and any issues from previous days — blockers or bottlenecks. From there it’s sometimes absolute chaos; I feel like I’m running in five directions with my patients, my staff, everything that goes on in a dental office. But man, the day flies by. Sergei, what do you see in the future for CAD/CAM and AI? I know you probably can’t give away the company secrets, but what does the future look like for us?

Sergei Azernikov: The future looks exciting. There’s a lot of promising technology. One thing I’m excited about is voice interfaces, which I think will play a very significant role in the dentist’s office. In the past there was some hesitation about privacy, but today I see more and more dentists using those tools, and there’s a lot to be done there. As it becomes smarter, more and more operations will be doable just by speaking — maybe to the point where you have a virtual assistant that does all your bookings and scheduling, and even potentially talks to your patients. All of that is actually doable today. So that will be a big change in the relatively near future. Another opportunity, which looks a bit futuristic today, is technology that may let you introduce more robotics into your practice. We’ve been looking at technologies where companies try to build a robot that can automatically prep, with the right imaging modality. That may be a bit far in terms of adoption — when will patients be ready to accept that a robot will prep them? But today I drove my Tesla — we talked about this when you were here — and I trust the full self-driving system now more than myself. So maybe we’re not that far from the day when a patient will trust a robotic doctor. It’s already happening in other areas of medicine, so I think it’s coming; it’s just a question of time. And if you think about software, what’s really missing in dentistry today is a platform. When you introduce a new system to your practice, you have to figure out how it’ll interplay with the other systems — there’s no gluing layer. Sometimes it gets to the point where you need IT personnel to take care of it, because it can be complex for many dentists. So that’s a huge opportunity, and it will change — I’m sure it will, because there’s a gap there. You need something similar to what Microsoft did with Windows, where any type of software, as long as it provides the interfaces, can be plug and play — or any USB device you plug in and it automatically finds the required drivers and works. That’s the situation I expect in dentistry. It’s something we’ve been thinking about, and it obviously requires collaboration between the significant players in that space. But that’s where I think the bigger shift will come — again, because development became easier with AI. Software development became much more streamlined, so it’s easier now to come up with those systems and interfaces. I think it’ll happen relatively soon.

Dr. Eric Block: That’s awesome. All right, Sergei, final question — I know you’re a busy man. What advice would you give to the young dentists out there, or young entrepreneurs in general?

Sergei Azernikov: Young dentists and young entrepreneurs may have different goals. For the young dentist — I was talking on my podcast to Jeremy Krell, one of my first guests, and I like the way he put it: you cannot afford to stay. If you come into a profession and think you’re just going to do what others did before, that’s not a good idea. You have to look around, and sometimes it’s difficult because there’s a lot of hype, but you really need to identify and think about technologies that are promising and genuinely help you, and try to adopt them. Because if you hesitate, others will run faster and the competition becomes stronger. This is especially true for doctors who want to stay independent. If they go to a DSO, it’s a different play, but if they want to stay independent, they need to be as effective as possible, making sure the practice runs efficiently using the tools available. That’s for the doctor. For the entrepreneur, it connects well to what I just said: there’s a huge opportunity in dentistry, because it was historically underserved versus other branches of medicine. There are lots of opportunities to improve interoperability between systems — the development of that platform. Once it happens, it opens the way for developers to get in. Think about the Apple App Store, or Google Android — it’s open to everybody. Imagine something like that in dentistry, where with a very low barrier somebody can develop applications available to doctors immediately. That’s the future I’m looking for. If you’re an entrepreneur looking to get into that space, that’s what you need to think about: how to get to the widest possible market and make it available quickly so people can utilize it. That, I believe, is the next big thing in the industry.

Dr. Eric Block: Sergei, thank you so much. That was such a great episode. Again, Sergei Azernikov, thank you so much for joining us.

Sergei Azernikov: Thank you so much, Eric. It was a great pleasure speaking with you, and I hope it’ll be an interesting episode for your viewers.

Dr. Eric Block: Thank you for tuning into the Stress-Free Dentist Podcast. If you enjoyed this episode, be sure to subscribe, leave a review, and share it with another dental professional who could use a little less stress in their life. And head over to thestressfreedentist.com to check out my books, resources, and upcoming events, all designed to help you find more balance and joy in your dental career. And remember, you’re not alone in this journey.