What You’ll Learn
- Why the marketing around AI is overhyped but the technology isn’t
- What percentage of patients ask for a human when an AI answers the phone
- The difference between open loop automation and closed loop automation
- Which front office tasks AI can actually handle beyond answering calls
- How to implement AI without throwing something at the wall and hoping it sticks
The Honest Answer Is Yes and No
Three years after ChatGPT launched, almost every dental vendor has slapped AI on their logo. Walk any show floor and you’ll hear the pitch over and over. AI-powered this. AI-enabled that.
Connor Ludlow has been building AI for dental practices since before the hype cycle started. As founder and CEO of Annie, he’s watched the conversation shift from curiosity to FOMO to something approaching exhaustion.
His answer to whether AI is overhyped? Yes and no.
“I would actually say yes is the main answer,” Connor told Adrian Lefler on a recent episode of the Byte Sized Podcast. “A big part of that is the hype word. It’s the way it often gets framed. You go to a show and sometimes it seems like every speaker is talking about AI and it comes across a little FOMO. If you’re not doing this, you’re missing out.”
The framing is the problem. When practices buy because something says AI, they end up with failed rollouts and burned bridges. When they start with a specific job that needs doing, the results look different.
“I don’t think I’m not bullish on it,” Connor said. “I think it’s an incredible technology. But the narrative around it can be difficult after three years.”
The Problem With Definitions
The AI conversation is distinctly frustrating because the term doesn’t mean anything specific anymore.
Before ChatGPT, there were clear categories. Machine learning. Statistics. Automations. Workflows. Google Maps rerouting you around traffic was one thing and a chatbot was another.
Now everything gets lumped together. A practice says they have AI and they might mean diagnostic imaging. They might mean an answering service. They might mean they use ChatGPT to write marketing copy.
“It’s like me coming to a party and saying I play sports,” Adrian said. “Okay, well, what sport? Backgammon? That’s not a sport.”
When a vendor says they’re AI-powered, it could mean anything. It could mean nothing. The label has become so broad that it’s essentially meaningless without specifics.
| What “AI-Powered” Could Mean | What It Actually Does |
| Diagnostic imaging tools | Analyzes X-rays for pathology |
| Phone answering service | Handles inbound calls conversationally |
| Marketing automation | Generates copy or schedules posts |
| Workflow triggers | Sends texts based on database events |
| Practice analytics | Surfaces patterns in production data |
Do Patients Hate It?
The most common question Annie gets from practices is, “Are my patients going to hate this?”
The concern is legitimate. We’ve all been trapped in bad phone trees. Press one for this. Press two for that. Say “schedule” and hear “oh, you want to pay a bill?” We’ve been trained to say “customer service representative” the second we hear an automated voice.
“Bad phone trees have trained us to hate these things,” Connor said. “If we had never had that experience, I think the discourse around these agents would be very different.”
The data tells an interesting story. When Annie launched in 2024, about 20% of callers asked for a human immediately, before the AI even said anything useful. Two years later, that number is down to 14%.
Still significant, but trending in the right direction.
“If the agent’s going to help me get whatever I need taken care of, I’m in on that,” Connor said. “I don’t think most patients would reject that. It’s just that we’ve been trained to correlate AI with not getting our objective taken care of.”
The numbers get even better in orthodontic and pediatric offices, where younger patients are more comfortable with chat and text.
Good AI vs. Bad AI
Adrian shared two recent experiences that captured the divide perfectly.
One call with a major company was seamless. Natural language. No phone tree. He explained what he needed and got routed instantly. Problem solved.
Another call was a disaster. The agent kept picking up background noise. The logic was still stuck in phone tree methodology. It glitched constantly. This was the most frustrating call he’d been on in months.
“The answer to me is patients are probably not going to resist AI receptionists as long as the freaking thing is really good,” Adrian said. “But if it’s not, it’s just a shitty technology and you’re frustrated with it.”
This is the real divide, between AI that works and AI that doesn’t.
Consumers accept technology that solves their problem and they revolt against technology that creates new ones.
Why Rollouts Fail
The number one problem with AI in dental practices isn’t the technology. It’s implementation.
“If you just look at AI as a silver bullet and throw something at the wall, odds are you’re probably not going to get the best results,” Connor said. “You’re maybe not even going to know what it’s actually doing, or you’re just going to see it fail once and burn a bridge.”
AI is technically software, but it behaves differently. There’s variability. Less structure. Unfamiliarity with how it responds.
Connor’s advice is to start using everyday tools like ChatGPT so the team learns how AI behaves. Get familiar with how it responds. Then pick one workflow with a clear trigger and see if the tool can actually do the job.
Annie assigns every new customer a human trainer who onboards the agent like a new employee. Because that’s what it takes. The practices that succeed treat implementation like hiring, not like installing software.
Open Loop vs. Closed Loop
So where does AI actually add value?
Traditional automation creates tasks for humans. Send an appointment confirmation text. Patient replies “can I reschedule?” That creates a task for someone to read the text, call the patient, and close the loop.
That’s open loop automation. It starts work but doesn’t finish it.
Closed loop automation actually completes the job. The AI reads the reply, understands the request, and reschedules the appointment. No human step required.
“A lot of automations create tasks for humans still,” Connor said. “Where we think AI is going is closed loop software. Something that’s actually going to end to end complete that task so that when you pull up Annie, it’s like sweet, Annie took care of that. Annie did the job.”
This is the real unlock. Not AI that creates more work for your team. AI that finishes work so your team doesn’t have to.
The Digital Coworker Concept
Annie started with website chat and missed call coverage. The product has evolved into something Connor now calls a digital coworker.
The idea is an always-on babysitter over the whole patient base. Not just handling calls when they come in. Watching every patient at every moment. Noticing when someone’s overdue. Recognizing when insurance information is missing. Seeing when an appointment opens up.
“What if you had a babysitter over your patient base?” Connor asked. “What if every single patient, you knew what was going on with them? If they’re overdue, if they’re up to date, if they’ve moved, if they have outstanding treatment.”
The capabilities now include:
Recare outreach. When a patient doesn’t schedule their next cleaning, Annie reaches out a month or two after the appointment. If they’re already overdue, she texts or calls to get them back on the schedule.
Reappointment. Similar to recare but proactive. Getting ahead of patients before they become overdue.
Appointment confirmations. Not just sending the text, but handling the replies. If someone says they can’t make it, Annie reschedules instead of creating a task.
Insurance verification. Getting insurance information from patients and putting it in the system.
Cancellation management. When an opening appears on the schedule, Annie reaches out to the waitlist to fill the slot.
“Those might not be as sexy,” Connor said. “But they become very feasible with the way AI operates.”
The Labor Shortage Connection
DSO executives and practice owners keep naming the same top complaint: recruiting clinicians and front desk staff is brutal.
And when you do find someone good, they get slammed with tasks. Recare calls. Confirmations. Chasing down insurance information. The busywork creates burnout. The burnout creates turnover. The turnover makes the shortage worse.
“That good employee gets slammed with tasks and is working a ton and that creates burnout,” Connor said. “And that keeps the flywheel going.”
Offloading repetitive work to an AI agent keeps good employees doing the high-value work they were hired for. Treatment coordination. Patient relationships. The things that actually require human judgment and emotional intelligence.
“That’s why we call Annie a digital coworker versus a digital employee,” Connor said. “We want her to feel like a sidekick. You can retain those employees that you do find because they can focus on doing their job and don’t get burnt out while doing it.”
What’s Coming Next
The roadmap follows the same logic. Find repetitive, monotonous tasks with clearly defined triggers. Build closed loop automation that finishes the job.
Lead follow-up is on the list. Patients who came in through the website but never scheduled. Form completion reminders. Following up when patients don’t fill out their paperwork before appointments.
Payments and overdue balances are coming. The technology exists to take PCI-compliant credit card payments over the phone. It just has to be built right.
Revenue cycle management gets requested constantly. Connor’s team focuses on front desk workflows for now, but the pattern applies anywhere there’s repetitive work that a human currently has to juggle.
“I think we’ll look back at something like recare and say, man, I can’t believe we were manually putting together a CSV and typing in every single phone number,” Connor said. “That’s where a lot of the burnout comes from. How many things you have to juggle and switch between.”
In This Episode:
Connor Ludlow, Founder and CEO of Annie
Conner Ludlow is the founder and CEO of Annie, an AI digital coworker built for dental practices. He is a software engineer and former CTO who previously led technology at a nationwide dental call center. He launched Annie in 2024 with website chat and an inbound AI receptionist, and has since expanded it into recare, reappointment, appointment confirmations, insurance verification, and cancellation management.
Adrian Lefler, CEO and Co-founder of My Social Practice
Adrian Lefler, CEO of My Social Practice, is a seasoned expert in the dental marketing industry with 14 years of experience. He is widely recognized for his engaging and informative presentations. Based in Suncrest, Utah, Adrian shares his life with his wife, four children, and a lively mix of pets. My Social Practice is a leading dental marketing company, and Adrian is passionate about helping dental professionals succeed in this dynamic field.
Frequently Asked Questions
Is AI in dentistry overhyped or worth adopting right now?
The marketing is overhyped. The technology isn’t. Buying because a product says AI leads to failed rollouts. Pick a task that’s repetitive, always on, and clearly triggered, then judge the tool by whether the job gets done. Practices that start with the problem instead of the acronym get results.
Why do AI rollouts fail in dental practices?
They fail when a practice throws AI at the wall with no defined job. Start by using everyday tools like ChatGPT so the team learns how AI responds. Then pick one workflow with a clear trigger. Annie assigns every new customer a human trainer who onboards the agent like a new employee.
What front office tasks can AI handle beyond answering the phone?
An AI receptionist can handle recare outreach, reappointment before patients go overdue, appointment confirmations, insurance verification, website chat, and cancellation management that offers open slots to the waitlist. The pattern is repetitive work with a clear trigger. Lead follow-up, incomplete forms, and overdue balances are next on the roadmap.
How many dental patients ask for a human when an AI answers?
About 14%, down from roughly 20% when Annie launched in 2024. That’s the share who request a human immediately, before the agent says anything useful. The number runs lower in orthodontic and pediatric offices, where younger patients are more comfortable with chat and text.
Can AI help with the dental staffing shortage?
Yes. Clinician and front desk recruiting is the top complaint from DSO executives and practice owners. The good hire you do land gets buried in busywork. Offloading recare calls and confirmations to an agent keeps that person doing the high-value work they were hired for, which is what keeps them.
