I have a hard time looking at something that's working, a clinic, a referral pathway, a piece of a business, and just leaving it alone. It's not that I don't recognize when something is functional. It's that "functional" and "as good as it could be" are two different standards, and I only ever seem to aim for the second one.
That instinct showed up early, in a much smaller form. As a spine surgeon, I noticed small frictions in hardware that had worked for years, and that noticing eventually led to two U.S. patents on spine implant design. But the pattern didn't stay confined to the operating room. It followed me into how I think about entire organizations, and it's the instinct that has done the most work in building LUX Healthcare.
Where the instinct did the most work: Lumin Health
Founding Lumin Health is where I first applied this at scale. It grew into a physician-driven, vertically integrated system, 7 urgent care and primary care clinics, 5 freestanding emergency centers, 2 imaging centers, 2 physical therapy and rehabilitation centers, multi-specialty surgical clinics, ambulatory surgery centers, and surgical hospitals. None of that happened as one finished plan. It happened because I kept treating each working piece as a draft, not a finished product: where a referral handoff was slower than it needed to be, where two departments were solving the same problem without talking to each other, where a process existed because it always had, not because anyone had recently asked if it still made sense.
I've written elsewhere about the mechanics of that, what actually separates good vertical integration from the version that just adds complexity (Vertical Integration in Healthcare), and what makes an MSO structure serve the physician and the patient instead of just the capital behind it (The Future of MSO Strategy). Both of those came directly out of this same habit: assume the current version is a draft, and go find the specific thing that's actually broken.
The same discipline, applied to LUX Healthcare today
LUX Healthcare's platform model, SpineLUX, MindLUX, NeuroLUX, ImagingLUX, and RegenLUX, exists for the same reason Lumin Health did. None of these specialties are a new idea. Interventional pain management, behavioral health, neurology, imaging, and regenerative medicine, delivered by our physician team, connected under one coordinated model, is a structure other people have tried before. What matters is refusing to accept the version of that structure that already exists elsewhere: the ones with weak governance, vague incentives, and systems that work fine until growth exposes exactly where they weren't actually built to last.
Every specialty platform we add gets held to the same test the pedicle screw modifications did years ago: is this a specific, measurable improvement to how care actually gets coordinated, or is it just a bigger version of the same structure? If it's not the former, it doesn't get built yet, no matter how attractive the opportunity looks on paper.
A platform is never finished. The moment you stop asking what isn't working yet, you've stopped building it and started defending it.
Where the risk actually is
The failure mode I watch for isn't complacency, it's the opposite: adding a new specialty or a new location because it's available, not because it closes an actual gap in how patients move through the system. Growth for its own sake is exactly the kind of "improvement" that makes a platform worse, not better. The discipline is the same one that kept the pedicle screw work useful instead of just novel: name specifically what isn't working, measure whether the change actually helped, and know when something genuinely doesn't need touching yet.
The same question shows up, more recently, in how I think about my own health, treating sleep, training, and recovery as a system worth auditing rather than something to just maintain. That's a smaller-scale version of the identical habit, and it's the subject of most of what I write now on longevity and performance.
Where this leaves the platforms now
ImagingLUX is still in development for exactly this reason, it doesn't launch until the coordination problem it's meant to solve is actually clear, not just because the category is available. That's the same standard every piece of Lumin Health, and now LUX Healthcare, has had to meet: not "does this work," but "is this actually as good as it could be, and what specifically would change if it were."