What 46 Orthodontic Practices Learned in Their First Month with AI
In early 2026, Heartwise Collective trained 46 orthodontic practices and over 105 individual participants through the AI Practice Advantage program, a six-week implementation course focused on building HIPAA-compliant AI systems for orthodontic operations. The cohort included practices ranging from under $1M to over $7M in annual production, with participants spanning every role: doctors, practice managers, financial coordinators, treatment coordinators, marketing coordinators, and clinical team members. This is what we learned about what actually happens when orthodontic practices start using AI.
I want to be upfront about something. Most of what you read about AI in orthodontics right now is either academic research about clinical AI (cephalometric analysis, aligner algorithms, imaging) or marketing copy from companies selling AI tools. Almost none of it comes from someone who has actually sat in the room with orthodontic teams while they build and implement AI systems for the first time.
That's exactly what we did. For six weeks, I watched over a hundred people learn, struggle, break through, and fundamentally change how they think about their daily work. Here's what happened.
Where Practices Were Starting From
We surveyed confidence levels at the beginning of the program. On a scale of 1 to 5, the majority of participants rated themselves between 2 and 3. Most had used only the free version of ChatGPT, primarily for personal tasks like travel recommendations, basic email drafts, and occasional social media ideas.
Almost nobody had used Claude, built an AI project, created a skill, or understood the difference between generative and agentic AI. They knew AI existed. They knew it was important. They didn't know where to start, and they were nervous about HIPAA.
That's important context, because what happened next was fast.
The Implementation Velocity Surprised Us
Week 1: Foundations and fear
Week 1 was orientation. We covered AI guardrails, HIPAA boundaries, account setup, and basic prompting frameworks. The energy in the room was a mix of excitement and genuine anxiety. People were worried about making mistakes, violating HIPAA, or feeling left behind by teammates who were moving faster.
The most important thing we did in Week 1 was set expectations: you're going to feel overwhelmed, and that's normal. Don't let it take you out. Stay in it.
Week 2: The SOP breakthrough
By Week 2, practices were recording themselves walking through procedures via Loom, taking the transcripts, and dropping them into the SOP Architect we built together. SOPs that had been on the back burner for years were being created in under 60 seconds.
One practice estimated they'd been putting off updating their protocols for over two years because of the perceived time commitment. They knocked out multiple SOPs in a single session. That was the first real mindset shift: the things you've been avoiding because you think they'll take forever can now be done in the margins of a normal workday.
Week 3: Content that actually sounds like you
Week 3 was the Brand Champion build, and this was where the energy changed from "I'm learning a tool" to "this is going to change how I operate." Practices built AI projects that knew their brand voice, their ideal patient, and their communication style. The output wasn't generic AI content. It sounded like them.
One practice owner told us the content coming out of his Brand Champion was better than anything his marketing agency was producing. He started questioning whether he still needed the agency. That's not a knock on marketing agencies. It's a reflection of how powerful it is to have AI that actually knows your practice versus a generalist agency that serves dozens of clients.
Week 4: Financial analysis in 10 minutes
By Week 4, we introduced Claude Cowork, the agentic AI layer that works directly with files on your desktop. Practices learned to drop bank statements and credit card statements into a folder and have AI produce a full P&L with overhead analysis, category breakdowns, fixed vs. variable expense classification, benchmark comparisons, and an interactive dashboard.
The reaction was consistent across the group: most practice owners don't have this level of financial visibility, and the ones who do only get it annually from their CPA. Now they can do it quarterly, monthly, whenever they want, in about 10 minutes.
The 5 Most Common Challenges
1. Token and usage limits
Claude Cowork was burning through tokens faster than expected during the program. This was partly a platform issue (Anthropic was aware and working on it) and partly because participants were running everything in the most powerful model (Opus 4.6) when the lighter model (Sonnet 4.6) would have been sufficient for many tasks. We started coaching people to use Sonnet for everyday tasks and reserve Opus for deep analysis.
2. Getting the brand voice right on the first try
About 30% of practices needed to refine their Brand Champion output after the initial build. The issue was usually that the brand voice documents they provided were too thin, or they hadn't included enough examples of how they actually communicate. We added a bonus session specifically for this, and the practices that went through the refinement process ended up with better output than the ones who were happy with the first version.
3. Different skill levels in the same room
With 105+ participants, we had people who were building advanced multi-project systems by Week 3 and people who were still getting comfortable with the difference between chat and projects. This is the reality of teaching AI to any group: the spread is wide. Our approach was to build step-by-step SOPs for everything, offer optional support sessions, and create a course AI agent that participants could ask questions to anytime.
4. Technology friction
Claude's interface changed during the program (HTML output became the default, which confused people). Some participants were on iPads instead of computers, which limited functionality. A few had account setup issues. Technology will always be the friction point, and having real-time support to troubleshoot these issues was critical.
5. The overwhelm is real, but it passes
The most common emotion in Weeks 1 and 2 was overwhelm. Learning AI is genuinely a firehose experience. The practices that pushed through it and stayed curious came out the other side with a completely different relationship to their work. The ones who let the overwhelm stop them at Week 2 missed the breakthroughs that happened in Weeks 3 and 4.
The Wins That Mattered Most
Beyond the specific tools and systems, the real transformation was in how people started thinking. These are the patterns we saw emerge:
"How long would this take without AI?" became the default question before starting any manual task.
Practices started reframing from "we need to hire for this" to "can we build an AI system for this?"
Fear of obsolescence shifted to excitement about becoming more powerful. One participant's husband told her she wasn't going to be obsolete, and she said, "That's refreshing. Good to hear."
Participants started texting themselves AI ideas at 2 AM. The creative energy was contagious.
Multiple participants said they didn't want the six weeks to end and asked about advanced offerings.
The specific wins ranged from small (branded patient handouts created in five minutes) to significant (a financial coordinator who had never used Claude building a complex Asana automation board for insurance billing in her first week). But the common thread was this: every practice found at least one thing that changed how they operate, and most found several.
What This Means for Orthodontics
The orthodontic industry's conversation about AI has been almost entirely focused on clinical applications: diagnostic imaging, treatment planning algorithms, aligner technology. That's important work, but it misses the operational reality of running a practice.
What we learned from training 46 practices is that the biggest AI opportunities in orthodontics right now are operational, not clinical. SOP documentation. Content creation. Financial analysis. Team training. Patient communications. Schedule optimization. These are the areas where AI delivers immediate, measurable ROI, and they don't require any clinical integration, CBCT data, or FDA clearance.
The practices that will have the biggest advantage over the next 2-3 years aren't the ones with the most advanced clinical AI. They're the ones whose teams know how to use AI as a daily operational tool. And right now, that number is vanishingly small. Roughly 84% of people have never touched AI. Only about 6% pay for it. Practices that are building AI systems today are in the top 1% of AI users globally.
That's not a marketing claim. That's just the math.
Frequently Asked Questions
What AI platform did you use?
The program is built primarily on Claude (Anthropic), using both the browser and desktop versions. We use Claude because of its strong privacy defaults, project architecture, skills system, and writing quality. We also introduce NotebookLM for team training applications and Perplexity for research tasks.
Is the program HIPAA-compliant?
Yes. HIPAA compliance is covered on Day 1. The systems we build do not process protected health information in the standard Claude environment. For any use case involving PHI, we teach participants about compliant platforms and proper de-identification techniques.
What size practices benefit most?
Practices of all sizes participated, from under $1M to over $7M in production. The systems work at any scale. Smaller practices often see the most dramatic efficiency gains because they have fewer people doing more tasks. Larger practices benefit from the systematization across multiple locations and roles.
How much time does implementation require?
The live sessions are 90 minutes per week, plus approximately 2-3 hours for implementation between sessions. The real time savings begin almost immediately. Practices reported saving hours per week on SOP creation, content production, and financial analysis within the first few weeks.
When is the next cohort?
The next AI Practice Advantage cohort is scheduled for September 2026. Enrollment details are available at heartwisecollective.com/ai-group-coaching.
Want to be in the next cohort? The AI Practice Advantage program trains your team to build HIPAA-compliant AI systems in 6 weeks. Learn more about AIPA or request the masterclass replay to see the systems in action.
About the Author: Lindsay Quinn is the CEO and Founder of Heartwise Collective, an orthodontic consulting firm specializing in financial systems audits, accounts receivable recovery, AI implementation, and fractional COO services. She has trained 46+ orthodontic practices in building HIPAA-compliant AI systems through her AI Practice Advantage program and has over 22 years of orthodontic industry experience.