All of my dental tech stack conversations naturally start with what PMS is being used, whether it’s standardized across locations, and then we move outward into the other products they use, what integrates, what doesn’t, and where opportunities for improvement or consolidation might exist.

Last month I wrote about shadow AI and what happens when tools move faster than the structure around them. The response to that article, along with what I’ve continued seeing since then, pushed me toward thinking about another part of that same conversation.

That part being what’s happening at the PMS level right now with some of the recent announcements around AI being built inside the PMS itself, not from external sources.

A lot of these announcements still aren’t very mainstream in practice yet, but quietly, the systems at the center of the dental technology stack are starting to adapt to the pace AI is moving.

The PMS has always had one boring yet vital job. Be the system of record, the lifeline of the dental practice. Manage the data. Let everything else plug in around it. (Well, everything they decide they'd let plug in or didn’t build themselves.)

These legacy platforms are now beginning to build the AI tools that up until now had only been developed outside of them. Building tools isn't new for these incumbents, but the potential impact some of these recent additions could have on the systems long term seemed worth writing about.

Scheduling agents. Voice AI. Revenue cycle automation. Capabilities that used to belong to point solutions are moving more native inside the system of record itself.

For DSO operators, these tools you've been paying for separately have started to show up as options inside your PMS. In theory, this could really simplify and consolidate things, which for operators could be a very good thing.

For point solution vendors however, if the system of record starts moving into your category, being integrated is no longer a protected position, not that it ever was in the first place.

For newer point solution integration players, some doors may be narrowing. Though open API and MCP models are also creating new entry points, which is worth paying special attention to.

There's small but strategic moves forming in how PMS platforms are approaching AI product development.

Product integration strategies themselves aren’t new. PMS platforms have always connected with outside products in areas they were a bit weaker in. What is happening now is that many of these platforms are starting to build the AI capabilities themselves instead of just integrating something already built.

The advantage of building natively is that the AI is operating closer to the workflow and underlying data structure inside the PMS. Confirmations post directly back to the schedule with no API needed. RCM flags show up inside the same interface the billing team already works from. Voice charting writes directly into the patient chart.

With more external or bolt-on integration approaches, you can still get strong capabilities, but in some cases the APIs and write-back functionality aren’t as deep. That can create more operational friction depending on the workflow. More places where data crosses a boundary you may not fully control.

Also, when the API doesn’t have all the details needed, or the information doesn’t write back cleanly, guess who usually gets blamed?

Not the PMS company. Usually the point solution.

If you read my last piece, you already know what that creates over time.

Based on my research and conversations across the market, here’s how some of the larger PMS platforms appear to be approaching AI right now.

Planet DDS launched AI agents directly inside Denticon earlier this year. A Confirmation Agent that handles outbound calls and reschedules on the spot. A Scheduling Agent that works recall and waitlists automatically. Their roadmap includes a Dental OS Desktop where DSOs can access core products, third-party integrations, and an agentic marketplace for bringing in AI agents on demand across specific practices.

Mike Huffaker, their CRO, said something on the DSO Technology Show recently that I think is worth sharing here:

Their play isn't to do everything but to be the engine everything else runs through.

CareStack is building on multiple fronts at the same time. The VoiceStack acquisition brought AI call handling natively into the platform. Their broader AI suite covers patient engagement, insurance verification, clinical documentation, and RCM. One DSO customer reportedly cut average callback time from over three hours to under two minutes. Their Overjet partnership extends into the clinical layer as part of what they're calling the Smart Dental Platform. They also have a data partnership utilizing OS Dental’s capabilities to look inside clinical and financial data.

Curve Dental launched Curve FLO AI last August, native by design. Pearl's Second Opinion built directly into imaging, voice charting through Bola AI, AI-powered insurance verification. No switching systems, no add-ons is how they position it.

Open Dental as usual takes a different approach entirely. No native AI development, but a fully open API and published schema. Pearl, Bola AI, AI receptionist tools all plug in cleanly. It's the mix-and-match option for operators who want best-in-breed and have the internal capacity to manage it. It also leaves room for newer players to come into the space, which is part of why that model stays relevant.

Henry Schein One announced something at ThriveLive that's worth understanding.

They opened what’s called an MCP layer inside Dentrix Ascend. MCP is basically a way for AI tools and agents to securely connect into the PMS and interact with the data inside it without needing a brand new custom integration every time.

In plain English, instead of manually pulling reports or clicking through dashboards, you can simply type something like “show me production reports for these locations,” “find patients with unscheduled treatment over $1,000,” or “which offices have the highest cancellation rates this month?” and the AI can access the system and return the information directly to you.

Think of it as a more standardized way for AI systems to work with the software instead of constantly building around it. In some ways it functions like its own product layer sitting on top of Ascend, built for where things are heading.

From what Dr. Ryan Hungate, Chief Strategy Officer at HS1 announced at Thrive Live, it works across three tiers. “Ask” lets any team member query clinical, financial, and operational data in plain language. “Orchestrate” puts pre-built agents to work on eligibility verification, image quality, claims reconciliation, recall gaps, and patient follow-up. “Build” lets practices and DSOs configure their own custom workflows, dashboards, and agents without waiting on a developer.

One DSO executive on stage at ThriveLive who tested the product described being in a room that morning where his team was just typing in plain language what they wanted to see, and it was building in real time. No developer or waiting months on a feature request.

The agentic AI wave is here regardless of what any PMS vendor does, HS1 included. Tools like Claude, ChatGPT, and purpose-built dental agents are going to want access to practice data. That's already starting to happen.

The question is whether it happens in a controlled, governed way or just around the edges of the system.

That governance piece matters more than I think most organizations fully realize yet.

Paul Blocchi from Black Talon recently described this evolving AI infrastructure as “secure doors and trusted hallways,” the idea that as AI agents and orchestration layers begin moving more fluidly across systems, organizations need stronger visibility, permissions, and governance around how data is accessed and shared.

By building the MCP layer themselves, HS1 is getting in front of that. They're building the door before someone else walks through the wall.

It also repositions them. Not just as a PMS but as the platform dental AI runs on.

The HS1 Ask tier is one of the clearest early examples of this in dental right now. Plain language, live production data, no report to run. You just ask it stuff and it gives you answers.

Here’s where I’m going to take this a bit more theoretical because in some ways, this move starts pointing toward a much bigger change.

The PMS may eventually become less of a visible application layer and more of an infrastructure and data layer sitting underneath everything. Something closer to what people in technology would call “headless.” The workflows, automations, and AI agents become the interface while the underlying system quietly powers everything in the background.

You can already see small signs of this starting to happen. When workflows, reporting, scheduling logic, patient communication, insurance verification, and operational tasks start getting handled through an orchestration layer sitting on top of the PMS, users begin interacting less with the actual software interface itself and more with the outcomes those systems produce.

That’s part of what makes the MCP discussion interesting.

The traditional PMS screen and navigation structure may still exist underneath it all, but increasingly the interaction layer starts shifting toward agents, workflows, prompts, automation, and conversational interfaces instead of users manually navigating software the same way they always have.

I had a conversation recently with John Schwartz, former Executive VP at Pearl and current Executive Director at Weave, around where some of this could eventually go.

One point he made that stuck with me was that as agentic workflows mature and start handling more of the orchestration between systems, the way DSOs think about PMS platforms could begin to shift. The underlying data layer still matters, but the workflow and interface layer sitting on top of it may become far more flexible over time.

The incumbents still have enormous advantages. But rapid shifts in technology also tend to create room for new approaches over time for those crazy enough to want to build a PMS.

This isn’t unique to dental practice management systems either. There’s already broader conversation happening around whether AI eventually changes the role of large CRM systems like Salesforce in a similar way. Not necessarily replacing the underlying data layer, but changing how users interact with it day to day.

Dental may not move as fast as broader technology markets, but the same pressure eventually starts to show up here too.

The caveat is that none of this works without clean, structured data underneath it. The platforms building that foundation now are the ones best positioned for where this goes.

The innovation in AI has started to show itself more clearly, especially over the last few months.

We’ve seen steady improvements from innovative companies, and now the systems of record are building their own versions or finding ways to bring those capabilities closer to the core platform.

In a lot of ways, they’re catching up.

That makes evaluation and implementation more important than ever.

If you’re trying to set up your DSO or practice group for long-term success, the backbone of the business has to be able to support where technology is going. That doesn’t mean every tool needs to be native, or that every point solution goes away. But it does mean your core systems need to support AI capabilities, data movement, workflow automation, and the way your teams actually operate.

As these systems become more agentic and AI-driven, there’s also an entirely new infrastructure and cost layer forming underneath it all. Processing, orchestration, token usage, data access, automation workflows.

Somebody is paying for all of that activity over time.

These legacy systems will build what they can, partner where it makes sense, and find smart ways to keep users inside their ecosystem. That’s not necessarily a bad thing. It just means operators need to ask better questions and think longer term.

Who can support your growth as you add locations?

Who can help improve same-store performance?

Who can keep up as AI becomes less of a feature and more of an operating layer inside the business?

That’s the conversation I think more DSOs need to be having.

Please reach out if you want to talk through your process of evaluating and implementing technology, I can help.

Schedule some time with me at this link.

The DSO Technology Show

Brian Colao and I sat down with Mike Huffaker, the CRO of Planet DDS to talk about the importance of scalable, secure, adaptable PMS solutions.

👉 Watch the episode here

DSO Compass Off Script

Clayton and I break down:
• why trust starts and ends with the doctors in a practice
• where implementations break down
• AI creating pressure without clear outcomes

👉 Listen on Spotify, Youtube & Apple Music

Truly appreciate you being here and making it to the end. Have ideas, questions, or tech you want me to explore? Just reply to this or book some time with me to discuss.

- Matt