Technology is getting introduced into dental organizations faster than most teams can realistically evaluate, implement or manage.
Tools like ChatGPT and Claude are helping individuals and teams do more and move faster day to day, whether that’s writing communication, cleaning up notes, analyzing contracts, or pulling information together that used to take hours to do.
Some teams are going a layer deeper, with internal workflows getting built, systems being connected, and in some cases teams starting to build their own tools through what many are calling “vibe coding.”
They are building software products to support their business in a way that was once only done by experienced software engineers.
That’s a massive shift in the makeup of who is building technology.
Another important shift that’s happening is how these tools are being used.
It wasn’t long ago that most of the interaction with AI was purely question-based. Writing something, summarizing something, and waiting for an answer.
Now it’s starting to move beyond that. Tools are beginning to take action, whether that’s handling communication, triggering workflows, or operating in the background across systems to complete actual tasks.
That shift from asking to doing is a major breakthrough that seems to have happened overnight, but to institutional investors this is the path they envisioned.
The amount of capital flowing into AI right now is impossible to ignore.
In 2025, AI companies captured 65.6% of all U.S. venture capital, about $222 billion out of $339 billion total. A year earlier it was closer to 47%, and a decade ago it was around 10%. On top of that, major technology companies are expected to invest roughly $650 billion into AI infrastructure in 2026, up from about $410 billion in 2025.
So this AI thing isn’t slowing down. It’s accelerating and is certainly not just a "bubble".
Inside the black box
There’s a term floating around you may have heard of. Shadow AI.
What is it?
It can be as simple as an office manager dropping patient or insurance information into a tool to get a faster answer, whether that’s writing a referral, reviewing an EOB, or cleaning up communication.
On the surface, it feels pretty harmless and effective.
As my friend Jena Taft, Senior Counsel at Dykema, described it while speaking at an AI event last year, the data goes into the “black box.” A place where it lives after it’s used.
We don’t know where it goes, how many systems it touches, or who ultimately has access to it.
And in many cases, it’s not just one system. A single request can move across multiple models and environments, where one processes it, another refines it, and it may be stored or backed up somewhere else entirely.
What feels like one interaction can end up touching multiple systems behind the scenes.
Andy Taylor, CIO at Children’s Surgery Center, also chimed in on this and put it simply:
Most organizations don’t have a full picture of how AI is being used across their teams. It’s not intentional, but it creates gaps that are difficult to close later.
This Shadow AI is not limited to the practice level either. The same patterns are showing up at higher levels inside organizations, where I’d argue the impact can be even more significant.
In some cases, sensitive operational data is being used to test and build workflows without much structure or security around it. It’s happening quickly, with tools that make it easy to experiment and get things done without waiting on bureaucracy.
These tools are useful, exciting, and reflect how fast things are moving, but they also introduce a different, and potentially more serious layer of exposure.
Regardless of where it’s happening, the pattern is the same. Things are moving faster than the structure around them.
Even when organizations want to get ahead of this, it’s not always straightforward.
I was in a conversation recently with an enterprise-level DSO executive and he stressed this multiple times:
Right now, there’s just more work than there are hands to do it.
These lean teams are trying to evaluate new tools while still managing everything already in place, and in some cases stepping into environments where not many people are fully familiar with this type of technology.
Even with the right intent, it’s not easy to keep pace with how quickly all of this is moving.
The AI graveyard
Most organizations have invested in AI tools over the last 12 to 18 months (if you haven’t, time to start), and what I’m seeing is that many of these tools aren’t even being used anymore.
Some didn’t integrate cleanly, adoption never fully happened, or they weren’t aligned to existing workflows in the first place. In other cases, security or data questions came up after the rollout and stalled the process.
So in plain English, they’re just sitting there. Not fully implemented, not fully retired. In purgatory.
If you haven’t seen this yet, there’s a good chance you will.
There’s also a broader exciting, yet scary shift happening.
In some ways, DSOs are starting to behave more like software companies, building internal tools, connecting systems, and creating workflows that sit between platforms.
In a number of cases, teams are being asked to build or connect systems internally without having true software development resources in place.
IT teams that have historically focused on infrastructure are now being pulled into building technology workflows and automation layers, which are very different skill sets.
What gets built often works for a period of time and solves an immediate need, but it tends to be fragile. Over time, those layers create their own form of challenges. When something changes upstream, whether it’s a platform update or a system change, those internal tools don’t always hold up
Gary Salman, Founder and CEO at Black Talon, put it to me directly:
The way we think about integrations is going to change. Agents will start doing the work. Platforms that don’t open up won’t keep pace.
As more workflows move toward automation and AI, closed systems are going to have a harder time keeping up. It doesn’t happen all at once, but the pressure builds over time.
There’s also a security layer that comes with this.
As more tools get introduced, especially ones that surface patient or operational data in their own interfaces, the number of access points increases. Each new surface creates another place where data can be exposed if it’s not properly controlled.
A lot of the conversation around AI right now are focused on what it can do. Less of it is focused on how it’s being used and how it’s being secured.
In further conversation with Gary he mentioned that at conferences like HIMSS, much of the discussion is already centered around secure AI adoption, data governance, and system design, which feels a few steps ahead of what’s currently happening in dental.
Where are we going?
The direction this is heading feels pretty clear.
As more workflows get built around automation and AI, systems that don’t open up are going to feel that pressure more over time.
Adapt or be Kodak.
But this isn’t just about keeping up with cool technology. Security has to be part of the equation, especially as more data moves across systems that aren’t always fully understood.
And just as important, better operational performance only happens if these tools are implemented the right way. Adoption, implementation, and how they actually show up inside workflows is what determines whether they make an impact or not.
That’s where this all leads. Not just more technology, but more thoughtful use of it.
Because at the end of the day, the reason to buy and use technology is to be more effective inside the business, not just to stay busy.
No one is building a tech stack just for the sake of it. With dental currently being so focused on operational efficiency, things need to work, and they need to work well.
All of this is worth paying closer attention to as it continues to evolve. And that evolution is moving quickly.
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.
www.dsocompass.com
Announcement….
The problem: buying dental tech is still a guessing game
Most dental tech decisions still come down to some combination of demos, referrals, and gut feel, not real data on what actually works.
So teams end up making high-stakes decisions based on vendor narratives and anecdotal feedback.
The result: tools that don’t integrate, slow or failed rollouts, and months of wasted operational time trying to make the wrong systems work.
I’ve been working on something behind the scenes that I’m ready to share.
It’s called DNTL Compass.
I’m consistently in conversations with DSOs and operators evaluating technology and the same gaps show up every time:
Limited time to evaluate, no clear framework for selection, and no trusted source of truth on what actually works once it’s implemented.
So decisions get rushed, shortcuts get taken, and teams are left dealing with the downstream impact.
DNTL Compass will be a platform for:
Verified user reviews
Side-by-side comparisons
Real-world insight into how technology performs
No pay-to-play. No vendor-driven rankings. Just signal.
If you’re evaluating technology, or trying to avoid another painful implementation, this will give you a faster, more reliable way to make decisions.
Less guesswork. Better adoption. Stronger outcomes.
Get early access to DNTL Compass:
👉 www.dntlcompass.com
Early contributors will shape what this becomes and get first visibility into what’s actually working across the market.
• Leave a review based on your experience
• Or sign up to be part of early access
New podcast episodes!
New episodes worth a listen.
Most dental conferences and technology conversations sound valuable but don’t actually translate into operational change.
In the latest DSO Compass Off Script, we break down:
• why some conferences drive real value and others don’t
• why operational strategy has to come before technology decisions
• and more….
👉 Listen on Spotify, Youtube & Apple Music
The DSO Technology Show
We sat down with the CEO of 3DISC to talk about where digital impression scanners are actually heading and what that means for operators making decisions today.
A bit of time to think
Was lucky enough to spend some time in Austin with a group I’m excited to work with.
Like most who’ve spent time on the road, there’s something about being in a hotel room with a view that slows things down a bit.
A little space to think, reflect, and reset.
I’ve found those moments tend to support what comes next.
Golf season is back
For anyone spending time in Utah, let’s tee it up.
I joined an indoor spot here in Park City trying to get my game back to where it was before the wrist situation in 2023.
Still a work in progress.
Always up for a round if you’re around, inside our out.
Also, Go Sabres! First playoffs in 15 years.
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
