AI in Practice: Driving Better Care and Better Economics — A Conversation with Two Founders

At Pearl Health, we spend a lot of time thinking about what it takes for provider organizations to thrive — not just clinically, but operationally and financially. And increasingly, the answer runs through artificial intelligence.

As Director of Growth Partnerships, part of my role is staying close to the companies and innovators building tools that can genuinely move the needle for the practices we work with. I'm constantly in conversations with founders, operators, and clinicians about what's working, what’s not, and what's worth paying attention to next.

But AI in healthcare can still feel abstract. The conversation often swings between breathless optimism and cautious skepticism, leaving practice leaders with a more practical question: what does this actually look like for my patients and my practice today?

That’s the question I wanted to explore—not from a 30,000-foot view, but from the people building these tools on the ground. To do that, I sat down with two founders approaching the AI opportunity from different angles:

1. Aaron Molitor, founder of Convoy Health, is building the financial operating layer for medical groups — using AI to replace fragmented, backward-looking reporting with real-time intelligence and actionable insights.

2. Luke Moretti, founder of AI Optics, is building AI-powered point-of-care retinal screening technology that enables primary care providers to detect vision-threatening diseases earlier and more consistently.

One is working on the clinical frontier; the other on the operational backbone. Together, they highlight how AI is reshaping healthcare from multiple directions at once.

What stood out in both conversations is that these companies are solving real problems—not introducing new complexity for already stretched teams. Their tools are designed to fit into existing workflows, not disrupt them. And in both cases, the value proposition is clear: better outcomes for patients and stronger economics for practices—something we believe at Pearl shouldn’t be a tradeoff.

What follows is a lightly edited, fireside-style Q&A covering the problems they’re solving, how their AI works in practice, what real-world adoption looks like, and where they see things heading next.

Part 1: Convoy Health

Aaron Molitor is the founder of Convoy Health, which is building the financial operating layer for medical groups. Convoy’s AI platform connects across EHR, billing, scheduling, and payroll systems to give practice leaders a unified, real-time view of their business—and the intelligence to act on it.

Ben: In plain language, what does your AI do that a traditional software tool simply couldn't?

Aaron: A traditional tool might show you that visits are down 8% this month. Our AI goes a layer deeper — it identifies that the drop is coming from two specific providers, traces it to increased no-show rates on certain appointment types, connects that to scheduling gaps earlier in the week, and recommends specific actions like adjusting scheduling templates or overbooking thresholds.

It's the difference between reporting and reasoning. Instead of giving you numbers to interpret, it gives you answers and next steps.

Ben: What does it feel like for a practice before and after adopting a system like this?

Aaron: Before, most practices were operating with a constant sense of friction — teams were pulling reports from different systems, reconciling numbers manually, and debating what's actually true. There's a lot of effort, but not a lot of clarity.

After, the shift is less about new features and more about confidence. Leaders can see a unified view of the business in one place, understand why things are happening, and act quickly. They can ask questions, get analysis, and pressure-test decisions in real time — like having a financial analyst available at all times. Instead of chasing data, they're making decisions. It turns a reactive, fragmented experience into something much more controlled and intentional.

Ben: What does the healthcare system get wrong today about practice finances?

Aaron: Practice finances are fragmented, reactive, and disconnected from operations. Decisions about staffing, scheduling, and patient access are often made without a clear financial lens — and finance teams are stuck looking backward instead of guiding the business forward. That disconnect leads to inefficiency across the system: underutilized providers, long patient wait times, and missed revenue opportunities.

At the surface, this looks like a finance problem — but it's really an access and care problem. When a practice doesn't have a clear view of its operations, patients wait longer and providers are underutilized. By tightening how a practice runs, you're not just improving margins — you're enabling more patients to be seen and improving the overall experience of care.

Ben: What's one thing you wish more practice owners understood about how AI can help them right now?

Aaron: There is a real, near-term competitive advantage to adopting AI early. Practices that implement AI today can unlock meaningful improvements in efficiency and EBITDA — from better utilization to reduced administrative overhead. Over time, those gains will get competed away as contracts, pricing, and industry dynamics adjust. Right now is the window where those improvements can be captured directly by the practice. Early adopters don't just improve operations — they get to keep the upside before the rest of the market catches up.

Ben: Where do you see your company, and AI in healthcare broadly, in five years?

Aaron: AI will become embedded in the background of nearly every healthcare workflow. What looks like slow tech adoption today actually creates the perfect setup — legacy systems and processes leave room for AI to layer on top and drive step-function improvements without requiring full system replacement. That creates an opportunity for entirely new categories of products that don't just digitize workflows, but fundamentally change how decisions get made.

For healthcare organizations, that means less administrative burden, more coordinated operations, and a shift from reactive to forward-looking management. Running a medical group becomes more structured, more efficient, and ultimately more scalable.

Part 2: AI Optics

Luke Moretti is the founder of AI Optics, which makes the FDA-cleared Sentinel Camera—a point-of-care retinal screening device designed for primary care settings. AI Optics is on a mission to bring retinal screening to the point of care at scale, so fewer patients lose vision from diseases that could be caught earlier.

Ben: What was the moment you realized point-of-care retinal screening was a problem worth solving?

Luke: Back in 2019, when I was a medical student at NYU, I had a firsthand experience with how hard it is to evaluate the retina at the point of care. The direct ophthalmoscope was difficult to use, and even when I could get a view, I was not confident ruling out retinal disease because I was not an eye specialist. Only eye specialists have seen enough retinas, both normal and diseased, to develop that confidence.

What stuck with me was the gap between how important evaluating the retina is and how inaccessible that evaluation still is in everyday care settings. I saw patients lose vision because they either did not have access to an eye specialist or they knew they should have gone and life got in the way. The disease then went undetected until it had progressed significantly. Coming from a medical device background, I started thinking about what kind of product could make retinal screening practical at the point of care. That was the moment the idea really clicked for what became our Sentinel Camera. With the help of an incredible team, we have turned that idea into reality and we're on a mission to bring retinal screening to the point of care at scale, so fewer patients fall through the cracks and lose vision from diseases we had every opportunity to catch earlier.

Ben: In plain language, what does your AI do with a retinal image that traditional software simply cannot do?

Luke: Our FDA-cleared Sentinel Camera makes point-of-care retinal screening practical today, and we are building toward future AI screening capabilities on top of that foundation. Traditional software can store an image, display it, or send it somewhere else. What it cannot do is meaningfully interpret what is in that image. That is the opportunity AI creates: a workflow where a provider can capture a retinal photo during a routine visit and AI can help flag the presence of disease and indicate which patients may need specialist follow-up. That is a fundamentally different model for screening because it brings retinal evaluation much closer to the patient, makes it much easier to do at scale, and gives you a real shot at catching disease earlier instead of later.

Ben: What does the healthcare system still get wrong about the relationship between better patient care and practice economics?

Luke: The uncomfortable truth is that even things that clearly improve care often do not get adopted if they are not financially feasible. Value is not only about quality in the abstract. It is quality relative to cost: Value = Quality ÷ Cost.

If you improve care quality but the cost goes up disproportionately, that is not actually higher-value care. What is exciting about our product is that it improves both sides of that equation. It helps practices improve care by increasing screening rates and enabling earlier detection of retinal disease, while also creating reimbursable revenue for the practice. That is why the value proposition is so strong. And the ultimate outcome is a healthier population with better vision to engage more fully with this wild world.

Ben: What gives providers confidence that an AI screening result can be trusted in real-world practice?

Luke: In healthcare, trust comes from rigorous validation. The bar has to be high. The technology has to be clinically evaluated, easy to use, and easy to understand, with a clear next step for the provider. Trust is foundational, but for real-world adoption the system also has to fit seamlessly into existing care workflows.

Ben: Where do you see your company, and AI in healthcare more broadly, in five years?

Luke: We believe point-of-care AI-based retinal screening will become standard practice, and AI Optics is building toward that future now. In five years, I expect our products to be deployed broadly across care settings, delivering millions of screenings and catching much more disease much earlier. If we do this right, we can start to bend the curve on vision loss from retinal disease, first slowing its growth, then flattening it, and ultimately driving it down. More broadly, I see AI transforming healthcare through two main avenues. The first is through existing software platforms: medical record integrations, workflow tools, note generation, and systems that can actually make sense of messy longitudinal patient data. The second, and the part I personally get most excited about, is physical AI in medicine:  where sensors capture real-world data and AI-based systems interpret those signals. A lot of medicine, especially the physical exam, is still a human being using one of the core human senses and asking their brain to interpret the signal. We diagnose by seeing, listening, touching, scanning, and measuring. Over time, more of those functions will be captured by digital sensors, and once you digitize those signals cleanly, you can apply deep learning and unlock entirely new ways to screen for disease earlier and more consistently. AI Optics fits squarely in that world. The retina is one of the most information-rich windows into human health, and we are building the tools to make evaluating it practical at the point of care. If we succeed, we will not just make screening more accessible and efficient or help prevent vision loss for millions of people. We think this is just the beginning of what becomes possible when you can capture images at the point of care and apply deep learning to them at scale.

Where This Is Heading — and Why It Matters 

If there’s a through-line across both conversations, it’s this: the practices that move early on AI won’t just operate more efficiently—they’ll be better positioned to deliver proactive, high-quality care that patients need and that value-based models reward.

At Pearl Health, that’s the future we’re building toward. We’ll continue bringing you conversations like this—with the builders, clinicians, and operators turning that future into reality.

Interested in learning more, connect with Ben Breuer — and follow Pearl Health — on LinkedIn.