I see a lot of founders get stars in their eyes when they talk about AI in healthcare. They come to me with a brilliant model, a slick deck, and a plan to "revolutionize" the industry. Most of them fail.
It’s not because their tech isn’t good. It’s because they don’t understand the brutal, non-negotiable realities of the healthcare market. After two exits of my own and investing in over 200 companies, including more than 20 in the AI healthcare space, I’ve seen firsthand what separates the winners from the well-intentioned failures. It’s almost never about the algorithm.
If you’re a founder in this space, forget the hype. Here’s the uncomfortable truth about what it really takes.
1. The Model is the Easy Part
This is the single biggest delusion I see. Founders spend 90% of their time perfecting a model to get a 98% accuracy rate instead of 97%. Guess what? No one cares. Your model is a table stake, not a business.
I once met a team from Stanford—absolute geniuses. They had developed a diagnostic AI that could spot a rare form of cancer from imaging with incredible precision. They thought the tech was so good it would sell itself. They were wrong. They couldn't get the data to train it at scale. They couldn't get it integrated into a single hospital's ancient EHR system. The company died a slow, painful death.
The real work, the real moat, is everything else: navigating the byzantine data privacy rules, integrating with a dozen different legacy systems, and surviving the regulatory gauntlet. Your brilliant algorithm is just the ticket to the game. The game is won in the trenches of implementation.
2. Your "Customer" is a Committee of 10 People
In SaaS, you sell to a user. In healthcare, you sell to a bureaucracy. The doctor who will use your tool is probably the least important person in the buying decision.
Your real customer is a shadowy committee that meets once a quarter. It includes the department head (who cares about clinical outcomes), the CFO (who only cares about cost savings), the IT director (who is terrified your software will break their system), the legal team (who sees risk everywhere), and the procurement officer (who gets a bonus for negotiating you down 15%).
I had an investment in a company with a fantastic tool for automating clinical trial paperwork. The doctors loved it. It saved them hours. But the deal got stuck in the hospital’s procurement department for 18 months. The startup nearly ran out of cash waiting. You have to map this entire political landscape from day one and find an internal champion who can walk your deal through the door. Without that champion, you're just another vendor in a pile of a hundred others.
3. Reimbursement is a Brutal, Non-Negotiable Gatekeeper
This is where most healthcare dreams go to die. If you don't have a clear answer to the question, "Who pays for this and how?", you don't have a business. You have a science project.
Founders love to say, "We'll figure out the reimbursement strategy later." That's like saying you'll build the engine after you've designed the car's paint job. In the US system, the only thing that matters is whether there is an existing CPT code (Current Procedural Terminology) that a hospital can use to bill for your service.
Creating a new CPT code is a multi-year, multi-million dollar lobbying effort. It's a nightmare. The smart play is to find an existing, established code you can attach your service to. If your tool helps a radiologist read a scan faster, you're not selling a new service; you're making an existing, billable service more efficient. That's a story a CFO can understand.
4. "Automation" is a Dirty Word
Never, ever walk into a hospital and tell doctors or nurses that your AI is going to "automate" their job. They will see you as a threat, and they will find a thousand reasons to kill your project.
It’s a simple psychological truth. No one wants to be told a machine can do their job better than them. The resistance isn't logical; it's emotional.
The right way to frame it is augmentation. Your AI is a "co-pilot" or an "assistant." It doesn't replace the clinician; it frees them from the soul-crushing administrative work that burns them out. Clinicians spend up to 40% of their day on paperwork and charting. Your AI does the grunt work so they can spend more time with patients. You’re not replacing them; you’re restoring them to their actual job.
5. The Data Moat is Real, But Not How You Think
Everyone talks about having a "data moat." Most of them are talking about a puddle. They think that just having a large dataset is a competitive advantage. It’s not.
First, most large datasets are messy, unstructured, and full of errors. Second, your competitors can often get access to the same or similar data. The real, defensible moat comes from a unique, proprietary feedback loop.
One of my most successful investments didn't have the biggest dataset. But they had an exclusive partnership with a network of specialty clinics. Their AI would make a recommendation, and a team of top specialists would review and correct it. That corrected data was fed back into the model, making it smarter with every single use. The AI was learning from the best doctors in the world, in real-time. No one else could replicate that data stream. That’s a moat.
6. The Best Founders are Clinician-Engineer Duos
I see two types of failed teams. The first is all engineers. They build technically elegant solutions to problems that don't exist in the real world. They don't understand the clinical workflow, the political realities, or the user's deep-seated habits.
The second is all clinicians. They have a fantastic idea but no clue how to build it. They can't separate what's technically possible from what's science fiction, and they can't manage a product roadmap.
Every single successful AI healthcare company I've backed has had a hybrid founding team. You need a doctor or a nurse who has lived the problem, who understands the workflow in their bones. And you need an engineer who can translate that messy reality into clean code. Without that combination of domain expertise and technical expertise at the very top, you are flying blind.
7. Forget "Disruption." Think "Integration."
Silicon Valley loves the word "disruption." In healthcare, it's poison. The healthcare system is a massive, entrenched, and incredibly complex organism. It is designed to resist change, often for good reason—people's lives are on the line.
You are not going to burn it all down. You are not going to replace the EHR. You are not going to fundamentally change the way a hospital is run in your first five years.
The winners don't try to disrupt. They integrate. They build tools that slot seamlessly into the existing, chaotic workflows. They make their software feel like a natural extension of what the clinician is already doing. The goal is to be so useful and so easy to adopt that you become part of the furniture. The revolution in healthcare won't be a violent overthrow. It will be a quiet, almost invisible integration into the tools and processes that already exist.
Investing in this space isn't for the faint of heart. It requires patience and a deep respect for the complexity of the industry. But for the founders who are willing to get their hands dirty and solve the real, unglamorous problems, the opportunity is immense. Just remember: it's not about the AI; it's about the healthcare.
Frequently Asked Questions
How do I know which items apply to my situation?
Start by honestly assessing where your biggest bottleneck is right now. The items that address that specific constraint will give you the highest return on your time and energy.
Which item on this list has the highest impact?
It depends on your stage and context, but in my experience, the items near the top of the list tend to have the broadest applicability. That said, sometimes the less obvious items create the biggest breakthroughs for specific situations.
Can I implement all of these at once?
I'd strongly recommend against it. Pick the 2-3 items that resonate most with your current situation and focus there. Trying to do everything simultaneously is a recipe for doing nothing well.