$50 million for an AI-native dental platform. AI analysis moves into CBCT workflows. Radiographic findings begin feeding directly into insurance documentation. And researchers are asking whether dentistry is ready for autonomous AI.
This edition also includes a new AAAI-D interview.
A lot happened between August 15 and September 15. Here are the developments worth knowing.
$50 Million Goes Into an AI-Native Dental Platform
Archy raised a $50 million Series C, led by JMI Equity, to expand its cloud dental practice-management platform. The company reports serving more than 1,000 dental practices and processing more than $300 million in payments annually.1
What makes the investment interesting is where Archy is putting AI. Its announced tools extend across claims and collections, SOAP notes, eligibility verification, scheduling, patient communication and practice analytics. Rather than selling AI as another application attached to the practice-management system, Archy is building it into the system itself.1
If AI increasingly resides inside the software responsible for records, scheduling, billing and communication, choosing an AI platform may eventually become inseparable from choosing a practice-management system.
Pearl Brings AI Into the CBCT Workflow
Pearl commercially launched Second Opinion 3D for general dental practices in the United States and Canada. The platform analyzes CBCT volumes and identifies structures including teeth, the mandible, maxillary complex, nasal space, airway and inferior alveolar nerve canal. Clinicians can also request reports from licensed oral and maxillofacial radiologists within the platform.2
One distinction matters: this was not a new September FDA clearance. Pearl states that the underlying technology received FDA clearance previously; September marked its commercial launch.2
CBCT presents a different challenge from 2D dental radiographs. A volume can contain hundreds of slices and anatomy well beyond the clinician’s original area of interest. AI entering that workflow raises questions about interpretation of the complete volume, incidental findings and, eventually, whether independently validated systems can move beyond anatomical identification toward pathology detection and quantitative assessment.
When an AI Finding Becomes Part of an Insurance Claim
Trust AI announced an integration between VELMENI’s radiographic AI and Isaac PracticeOS. According to the companies, once a dentist confirms an AI finding, the system can identify required images, assemble payer-specific documentation, associate supporting radiographs and prepare a claim for review. A licensed dentist and billing specialist review claims before submission.3
The interesting part is what happens to the information.
A radiographic finding can move from image analysis -> dentist confirmation -> documentation -> claim preparation without each step existing in a separate system.3
That may reduce administrative work, but it also creates another place to examine errors: whether information retains the same meaning as it moves from a clinical observation into an administrative action.
“Agentic Dentistry” Enters the Literature
The phrase “agentic dentistry” now has a place in the scientific literature.4
Mohammed Turky and Paul M. H. Dummer published Are We Prepared for Agentic Dentistry? Exploring the Impact of Autonomous Artificial Intelligence in Dental Care in the International Dental Journal.4
The authors distinguish these systems from the task-specific AI familiar to dentistry today. Instead of analyzing one image or producing one output, agentic systems could potentially integrate information over time, coordinate tasks, support treatment planning and participate in patient follow-up.4
The terminology is worth watching. Dental software companies are already describing “agents” performing administrative work. The scientific literature is now considering what greater autonomy could mean clinically-including when a system should stop, when clinician input is required, how its actions are recorded and how responsibility is assigned.
$3.1 Million to Study AI Screening Outside the Dental Office
The University of Rochester’s Eastman Institute for Oral Health received a five-year, $3.1 million NIDCR grant to advance SMARTeeth++, an AI-supported system intended to identify caries and early periodontal disease outside conventional dental settings.5
Led by Jin Xiao, DDS, PhD, the project combines portable intraoral imaging, patient health information and AI to support screening, risk assessment and referral.5
Most dental AI begins after a patient has already reached a dental office and diagnostic images have been obtained. SMARTeeth++ is investigating whether part of that process can occur elsewhere.
The five-year grant is for research and development. It should not be interpreted as evidence that the system is currently validated for independent diagnosis.5
Also Worth Knowing
Can Dentists Build Their Own AI?
An International Dental Journal editorial asks exactly that question. No-code and low-code development tools are making it easier for clinicians without traditional programming backgrounds to create AI prototypes.6
But building a model and demonstrating that it is safe and clinically generalizable are different tasks. Dataset quality, external validation, privacy, cybersecurity and regulatory requirements do not disappear because software becomes easier to build.6
Prosthodontics Gets an AI Reality Check
A new Australian Dental Journal review examines AI across CAD/CAM, restoration design, decision support, modeling and patient communication. It also discusses automation bias, explainability, data protection, equity and medico-legal responsibility.7
For a specialty already built around highly digital workflows, the distinction between an impressive output and a clinically demonstrated benefit is particularly relevant.7
A Review of the Reviews Finds a Familiar Problem
An umbrella systematic review spanning radiology, periodontics, implant dentistry, caries, endodontics, prosthodontics and other fields found considerable potential for AI, while emphasizing external validation, generalizability, explainability and appropriate regulatory frameworks.8
For anyone evaluating an AI system, a practical question follows: Where else has this model been tested, on whom, and using what equipment?8
AI Takes on Credentialing and Revenue Cycle
Lightspun announced an AI-supported credentialing deployment with Specialty Dental Brands, a network of roughly 300 providers across 25 states. SDB reported reducing provider onboarding from as long as 12 weeks to less than two weeks. The result is company-reported rather than independently evaluated.9
Trust AI separately launched an RCM Hub in which software prepares claims, retrieves attachments and generates payer-specific narratives while dentists and specialist billers retain review roles.10
Neither application involves diagnosing a patient. Both show how much current dental AI activity is occurring in the work surrounding clinical care.
In Conversation
A Prosthodontist’s Perspective on Artificial Intelligence
Radek Mounajjed on evidence, clinical judgment, education, and the changing role of AI in dentistry
Interview by Carol Yassa
Interview Correspondent, American Academy of Artificial Intelligence in Dentistry®
About Radek Mounajjed
Radek Mounajjed, DDS, Ph.D., is an Associate Professor at Palacký University Olomouc, Institute of Dentistry and Oral Sciences, Faculty of Medicine and Dentistry, and Head Clinician at D.C.M Multidisciplinary Dental Clinic in Hradec Králové, Czech Republic. He is a Fellow of the Academy of Prosthodontics. His work combines clinical practice with academic teaching and research in prosthodontics and adhesive dentistry.
In this interview, Mounajjed discusses artificial intelligence from that perspective: as a practicing clinician and educator considering how these tools affect evidence, clinical reasoning, professional education, and patient care.
In your view, what distinguishes meaningful innovation in dentistry from transient trends?
I don’t think we can always clearly separate a short-term trend from a truly meaningful innovation. Many major innovations actually start out as small trends — as an idea or a tool that is still looking for its place. Over time, every micro-trend can either develop into a structured, clinically valuable tool, or simply disappear.
It’s a bit like a snowball: it either keeps rolling, gaining mass and direction, or it hits a tree and falls apart. We saw this, for example, with digital smile design and tools for predicting treatment outcomes. In the beginning, it often felt more like a “show” for the patient — a flashy preview of what the result could look like. Today, however, these tools are an integral part of esthetic planning and interdisciplinary collaboration.
In my view, real innovation reveals itself only over time. Not by how loudly it is presented, but by whether it actually improves the quality of care, sharpens decision-making, and reduces errors.
That is why I consider a true innovation to be any tool that can lift the level of care of an average dentist closer to the level of an expert. If a technology helps the clinician plan better, understand context better, and make better clinical decisions, then it is no longer just a trend. It is an innovation with the potential to change the whole field.
In your own clinical workflow, has AI meaningfully influenced diagnostic or surgical decisions, or is it still largely confirmatory?
Formally, we still divide AI into a confirmatory tool and a decision-making tool. Simply put: either I hold the wheel, or AI holds it. But in reality, it is not black and white. It is a gray scale that, in any given situation, leans more toward one side or the other.
A confirmatory tool, for me, is a situation where I already have an opinion and AI helps me verify my thinking. In my workflow, however, AI more often plays the role of an alerting tool. It brings a different perspective, widens my field of view, and points out connections I might overlook. Let’s say I am planning a clinical procedure and AI gives me a checklist of eight important points. Three of them I know well, three I am not entirely sure about, and two I am hearing for the first time. If those points are supported by sources and I can confirm they are not a hallucination, AI immediately moves me several steps forward in the quality of care.
This is how I see current AI most often. Not as a system that decides for me, but as a tool that alerts me, helps me verify my own thinking, and lets me look at a problem from a different angle.
Fully autonomous AI, where a clinician is no longer needed, is in my view still a thing of the future. I do worry, however, that this future may not be as far away as we think today — and the idea that one day an algorithm might make a clinical decision with no clinician above it honestly scares me a little.
When working with students and residents, where does AI genuinely add value, and where do you find yourself intentionally limiting its use?
When working with students and residents, I place the greatest emphasis on one thing: the ability to tell what is truly evidence-based from what is only feeling-based — an opinion, an impression, a tradition, or an attractively presented claim without a solid professional basis. This is exactly where AI offers enormous added value. It lets us quickly verify our own procedures and clinical habits. At the same time, it guides students toward structured thinking. And that is essential in medicine — in my view, one of the most dangerous doctors is a chaotic doctor. AI, when used properly, can help organize this kind of thinking.
That is why I encourage my students to use AI regularly. I see it as an exceptionally powerful tool that can develop their clinical reasoning and help them ask better questions. But AI must not replace their thinking. It should develop it, cultivate it, and confront it with evidence. Here I am very strict. I am allergic to the misuse of AI in the sense of: “I switch off my own brain and switch on AI.” That is the road to hell. If AI becomes a replacement for critical thinking instead of strengthening it, we are not producing better doctors — we are producing passive consumers of answers. And if society as a whole moved in that direction, we would risk a civilization without critical thinking.
On top of that, in dentistry there is a real problem with the fact that many clinicians love instant education — quick information, simple recipes, attractive claims. Because of this, part of professional education is shifting to social media, where claims are often accepted without being verified. And in medicine, we have not only scientists and genuine experts, but also scientific populists who can present unverified conclusions very persuasively.
And this is exactly where AI can play an extremely important role — not as a generator of opinions, but as a professional research and verification tool. To help clinicians, students, and residents navigate the literature faster, separate claims from evidence, and turn a chaotic amount of information into a structured, clinically usable framework.
As AI tools become more widely available, where are you actually seeing them used effectively in training today, and where are they still falling short?
The mere presence of AI does not automatically mean a good literature search or a clinically safe interpretation of results. AI can be an extremely powerful tool, but only if it is built on solid infrastructure, deep knowledge of the field, and well-designed prompts that respect the logic of the discipline. Today, I see very effective use of AI in areas such as digital planning, monitoring of manufacturing processes, and optimization of laboratory procedures. In those areas, AI already works very well, because it operates within relatively well-defined systems with clear parameters.
Where we are still lagging behind is the verification of facts, the interpretation of scientific literature, and the transfer of research results into clinical reality. A big problem is that publication in a high-impact journal does not automatically mean high-quality science. Even such a study may have methodological weaknesses, limited clinical applicability, or poorly interpreted conclusions. It is not enough to simply find studies — they have to be read and evaluated critically. This is where I see one of the greatest potentials of AI in education: not just to search the literature, but to help evaluate the quality of scientific work, find gaps in interpretation, and generate a pool of questions that have not yet been explored. In other words, AI should not just answer what we already know — it should help us better formulate what we still do not know.
Dentistry is a discipline that depends on a combination of several factors: skill, material, patient, and evidence-based dentistry. The outcome is not determined by the diagnosis alone. It is shaped by the clinician’s skill, the material used, the patient’s biological response, and the quality of the available evidence. That is why we need AI models that can better interpret, for example, in vitro research in relation to clinical parameters. It is not enough to say that a material had higher bond strength in a laboratory test. We have to ask what that means clinically, in what context, and for which patient.
Looking forward, I also see a path in closed systems, where an “army” of specialized AI tools will check medical records, documentation, the sequence of treatments, and the quality of clinical procedures. Not as a replacement for the clinician, but as an intelligent control layer above everyday practice.
What limitations in current AI tools are most apparent to you in day-to-day clinical or educational use that may not be obvious from published studies?
What bothers me personally the most is the inconsistency of outputs. The same AI tool, even with the same or a very similar prompt, will not always give the same answer. I see it a bit like a school of fish in the sea. AI casts a net — and sometimes it catches more good fish, sometimes fewer. Sometimes it finds exactly what we need; other times it misses part of the important information. The quality of the output always depends on the quality of the data being analyzed, but the amount of data today is so vast that the result is not always equally deep and consistent.
On top of that, there is the combination of what the tool retrieves at that moment and what it has learned in its own “black box.” That combination determines the quality of the answer, but it also explains why it is not always fully reproducible.
I am not an IT programmer. I am a dentist who works in a team with excellent programmers, and together we try to build the logic of the analysis as well as we can. It really is an incredible ride.
That is why I personally believe much more in highly specialized AI agents for specific problems. For professionals, a general AI chat cannot work in the long run as a universal solution. It is a bit like putting five kitchen robots into a single appliance — you may end up with an interesting hobby device, but not a professional blender. Professional medicine needs precision, specialization, and reproducibility.
From a research standpoint, what kinds of evidence or study designs are still missing before AI can meaningfully influence education or standards of care?
The dental industry today is pushing out an enormous number of new materials and procedures, often without sufficient real-world clinical testing. We have a lot of laboratory data, a lot of marketing claims, and a lot of short-term studies, but we still lack a robust link between in vitro research and real clinical outcomes from everyday practice. We need more high-quality clinical studies that logically build on in vitro data and rest on transparent methodology. And this is exactly where AI can play a major role — not by replacing science, but by watching over its structure, methodology, and real clinical relevance.
My personal vision goes even further. Imagine that AI could anonymously analyze clinical records in dental offices, track which materials and procedures are actually being used, for which indications, and with what outcomes. Legally and ethically, this is very difficult to do today, but philosophically, I see it as a fundamental direction. Every practice could in this way become a small scientific unit — not just a place where care is delivered, but also a source of anonymized, clinically relevant data from real life.
Thanks to this, we could track the impact of materials and procedures in real time — not only under the ideal conditions of a laboratory or a university study, but in ordinary, everyday practice. That is a huge paradigm shift. Today, we wait for years before a study appears that evaluates a particular material. In the future, AI could help create a continuous, anonymized, living body of evidence drawn directly from clinical reality.
In my view, the classic literature review, in its current form, is on its way out. This kind of work will be increasingly automated. But the future will not be about AI simply summarizing existing articles. The future will be about AI connecting literature, methodology, laboratory data, and real clinical data into a single intelligent system of knowledge.
As more AI tools enter the market, what should educators and institutions be most cautious about when introducing them into training environments?
What will matter is not just the adoption of AI tools, but above all the evaluation of the quality of that adoption. It is not enough to say that an institution uses AI. What matters is what kind of tool it uses, what kind of outputs it generates, who checks them, and what impact AI has on the way students think.
In the first stage, we will need expert bodies for individual disciplines — for example, a Scientific AI Board. Such a body should define standards and limits of use, and regularly evaluate the outputs of AI tools.
An AI product must not pass only on a technological or marketing level. It has to pass on a professional, clinical level. At the same time, we will need to build tools that check other AI tools — something like an AI auditing system. Human-issued certificates alone will soon no longer be enough, because AI is evolving too fast. It is a bit like trying to monitor a spacecraft heading to another galaxy using a telescope from a suburban observatory. The old control mechanisms simply will not keep up with the new speed of technology. For this reason, institutions should introduce AI not as a fashionable add-on, but as a professionally guided system with clear control over quality, safety, and real educational value.
Is there a risk that AI adoption may concentrate in certain practice settings and widen gaps in access to care?
I don’t fully share this concern. It may be partly justified in this early stage of AI adoption, when new technologies are naturally used mainly by more progressive practices and individuals. But something similar happened when computer systems were first introduced into dental offices. At the beginning, patient records were still written by hand, and digital documentation looked like something unnecessarily complicated. Today, it is almost impossible to imagine everyday practice without it.
It was the same with the automobile. At the start, it was a technology for progressive individuals, while most of society still rode horses. Today, it is hard to imagine riding a horse to work.
That is why I do not believe AI will remain, in the long run, the privilege of a particular group of practices. On the contrary — it will gradually become a normal part of dental practice. AI is no longer just an option or a technological curiosity. It is becoming a necessity for efficiency, quality of care, and the ability to keep up with the amount of information that modern medicine generates.
How do you balance innovation with professional responsibility when evaluating and adopting new technologies?
Every technology I bring into my clinical practice has to serve one main purpose: to improve the quality of care. Or, at the very least, I have to sense from the start that it has that potential. This is not just about direct clinical technologies, such as AI in intraoral scanners. Tools that handle administrative and repetitive tasks for me can be just as important. If AI saves me time, it means I have more room for the patient — and in the end, that is also an improvement in the quality of care.
When I evaluate a new technology, I often do a simple SWOT analysis. I like it as a quick way to think about the next step: what the technology offers, what its risks are, where its potential lies, and where it may turn into a dead end.
We have already seen a few such dead ends in dentistry. I remember, for example, electroanesthesia. The patient could increase the intensity of the impulse themselves, but there were so many wires around that I could hardly even reach inside the mouth. It wasn’t dangerous, but its practical value turned out to be very limited. On the other hand, there are technologies that can potentially be dangerous. In my experience, the first programs for reading X-ray images were dramatically poor and sometimes gave wrong information. Without sufficient expertise on the clinician’s side, such an output could harm the patient.
That is why I feel comfortable today as long as AI stays in the role of a support tool — a tool that supports and helps us, but does not decide for us. There I see plenty of room and a relatively safe framework for its use. But the moment AI starts moving from support into actual decision-making, we will need a clear legislative framework, a properly discussed code of ethics, and clear rules of responsibility. Innovation is necessary, but in medicine it must never run faster than our ability to take responsibility for it.
Looking ahead, how do you see AI realistically being integrated into dental education and clinical workflows over the next few years?
I see it as a certain reality. AI is already gradually being integrated into both education and everyday clinical workflows, and this process is going to accelerate quickly. One of the main roles of AI will be in the popularization of science — not only for students, but also for non-academic clinicians and everyday practitioners. AI can take a very complex topic and explain it so simply that even someone without a deep professional background can understand it. In my view, a good teacher is the one who can explain a complicated subject with perspective, structure, and simplicity. And AI does that very well. In many ways, AI is already today an excellent teacher.
At the same time, we need more sophisticated and specialized AI tools. We cannot rely only on general chat environments, because there is still the risk of hallucinations and a loss of professional context. I see AI as a bridge between dry science and practical everyday dentistry. I really enjoy teaching, and I always try to explain things to my students in a simple and understandable way. At the same time, I know very well that I cannot explain a problem properly if I do not understand it myself. AI pushes us to structure our thinking, verify the facts, and look for clear language.
I also see great potential in the fairness of education. AI can measure everyone with the same yardstick. If students perceive the educational system as fairer and more transparent, it can motivate them to work harder and at the same time reduce the room for illegitimate shortcuts.
AI-powered devices, trainers, and simulators will be a wonderful addition to education. But we must not forget one thing: at the end of the day, we will still be treating a living person. We have to prepare for a real patient, not just for a robot or a simulator. Technology can train us excellently, but empathy, responsibility, communication, and clinical judgment must still rest on a human being.
EDITOR’S NOTE
Responses have been edited for clarity and readability.
Around the Academy
Last Call for AAAI-D 2027 Abstracts
The submission deadline for the inaugural AAAI-D Annual Meeting is September 30, 2026.
The meeting, From Innovation to Implementation, will be held July 9, 2027, at UCLA Covel Commons in Los Angeles. Researchers, clinicians, educators, students and residents are invited to submit work for consideration.
Abstract submissions close September 30.
Submit an AbstractBuffalo Is Next
AAAI-D will be at the Buffalo Niagara Dental Meeting on November 6, 2026, with a morning educational program on artificial intelligence in dentistry followed by an afternoon Educators Roundtable for dental faculty.
The Roundtable will give educators an opportunity to compare how AI is being approached across dental education-including what institutions are teaching now, what they are considering next and where evidence remains limited.
Buffalo Program InformationLos Angeles 2027 Registration Is Open
Registration is open for the first AAAI-D Annual Meeting, July 9, 2027, at UCLA.
The one-day meeting will include invited speakers, scientific presentations, selected abstracts and discussions spanning clinical applications, research, education and implementation.
Early registration is available through January 31, 2027.
2027 Annual Meeting & RegistrationFellowship Applications Are Open
Applications are being accepted for Fellowship of the American Academy of Artificial Intelligence in Dentistry (FAAAI-D), the Academy’s highest professional distinction.
Fellowship is available through the Academy’s published pathways and applications are considered in quarterly review cycles. Applicants must be AAAI-D members in good standing and satisfy the requirements of the applicable Fellowship pathway. Fellowship is conferred following a structured peer-review process.
Fellowship Criteria, Pathways and Application InformationElsewhere in Dental AI
VA Holds Second International Symposium on AI in Dentistry
On September 18, VA Greater Los Angeles Healthcare System held its second international symposium on artificial intelligence in dentistry, with clinicians, researchers, educators and other experts participating from the United States and internationally.
The program covered diagnostic imaging, remote monitoring, clinical workflows, dental education, and applications of AI in large healthcare systems and military dentistry. Speakers also addressed evidence behind current applications, limitations of available systems, automation bias and questions of responsibility when AI is incorporated into clinical decisions.
The symposium was organized by VA Greater Los Angeles Dental Service in collaboration with VHA SimLEARN and ILEAD. AAAI-D recognized the symposium as an educational activity but was not an organizer of the event.
References
1. Archy. Archy Raises $50M Series C to Bring Dental Practices into the AI Era. September 11, 2026. Source
2. Pearl. Pearl Launches Second Opinion 3D, Bringing FDA-Cleared CBCT Imaging AI to General Dentists for the First Time. September 3, 2026. Source
3. Trust AI. Trust AI Just Made Your X-Ray Fight the Insurance Company. September 8, 2026. Source
4. Turky M, Dummer PMH. Are We Prepared for Agentic Dentistry? Exploring the Impact of Autonomous Artificial Intelligence in Dental Care. International Dental Journal. 2026. Source
5. University of Rochester Medical Center. EIOH Awarded $3.1 Million to Advance AI-Powered Oral Disease Detection. August 31, 2026. Source
6. Samaranayake LP, Tuygunov N, Osathanon T. Breaking the Code: Can Dentists Build Their Own Artificial Intelligence? International Dental Journal. 2026. Source
7. Faure M, et al. Artificial Intelligence Systems in Prosthodontics: Current Applications and Challenges. Australian Dental Journal. 2026. Source
8. Artificial intelligence application in dentistry: umbrella systematic review. Dental Journal (Majalah Kedokteran Gigi). 2026. Source
9. Specialty Dental Brands and Lightspun. SDB Partners With Lightspun to Automate Provider Onboarding and Credentialing. September 15, 2026. Source
10. Trust AI. Trust AI’s $2 Million RCM Hub Gives Solo Practices a DSO Back Office for $299 a Month. September 1, 2026. Source