I’ve been in the Silicon Valley game for a while now. Over a decade. I’ve seen a lot, built a couple of companies that got acquired, and now I spend a good chunk of my time as an angel investor. I’ve written over 180 checks, and a surprising number of them—more than 20—have gone to AI healthcare startups.
People ask me why. Why healthcare? It’s a notoriously brutal industry. It’s slow, it’s heavily regulated, and the sales cycles are enough to make even the most patient founder want to tear their hair out. They’re not wrong. It’s all true. But I keep coming back to it. Because the problems are real, and the impact—if you can get it right—is massive.
I’ve seen founders with brilliant AI models fail because they couldn’t get their foot in the door of a single hospital. I’ve also seen seemingly simple ideas take off because they solved a real, painful problem for doctors or patients. After watching more than 20 of these companies navigate the maze, I’ve learned a few things. Here are seven of them.
1. Your Model is Not Your Moat
This is probably the hardest pill for technical founders to swallow. You’ve spent months, maybe years, building a beautiful, elegant AI model that can predict a specific type of cancer with 99.9% accuracy. You think you’re done. You think the world will beat a path to your door.
It won’t.
In healthcare, the model is just the ticket to the game. It’s not the game itself. I saw this firsthand with a company I invested in that was working on an AI diagnostic tool for a rare pediatric disease. Their model was incredible. Truly groundbreaking. But they spent so much time perfecting it that they forgot about everything else. They didn’t think about how it would integrate with existing hospital EMRs. They didn’t have a good answer for how they would get it reimbursed by insurance. They had a great model, but they didn’t have a business.
Another startup in my portfolio, one that’s doing quite well, took the opposite approach. Their initial model was good, not great. But they focused on the workflow. They made it ridiculously easy for doctors to use their tool. It plugged right into the systems they were already using. The doctors loved it. And as they used it, they generated more data, which the company used to improve their model. They built their moat, not with the initial algorithm, but with the distribution and the data.
2. “Doctor in the Loop” is Not a Crutch, It’s a Strategy
There’s a certain arrogance that can come with building AI. A belief that the machine is, or will soon be, better than the human. In some narrow domains, that might even be true. But in healthcare, taking the human out of the loop is a recipe for disaster. Not just because of regulations, but because of trust.
I’ve seen founders pitch me on AI that will replace radiologists. I always pass. The founders who get my money are the ones who are building tools to augment radiologists, to make them better, faster, and more accurate. They’re the ones who understand that the doctor is not a bug, but a feature.
One of my most successful AI healthcare investments is a company that helps oncologists create personalized treatment plans. The AI analyzes all the latest research, clinical trials, and patient data, and then presents a set of recommendations to the oncologist. The doctor makes the final call. The AI is a super-powered assistant, not a replacement. That’s the right way to think about it.
3. Data is Everything, and It’s a Mess
This is true for all of AI, but it’s especially true in healthcare. The data is siloed. It’s unstructured. It’s full of errors. It’s a nightmare. And if you don’t have a strategy for dealing with that nightmare, you’re dead on arrival.
I once had a team pitch me on an AI for mental health. They had a great idea, but when I asked them about their data strategy, they just shrugged. “We’ll get it from the hospitals,” they said. I wished them luck. They didn’t get my money.
Getting data from hospitals is a long, painful process. You need to navigate IRBs, data-sharing agreements, and a whole lot of bureaucracy. The successful companies I’ve seen have a clever way around this. They either partner with a large health system that has a vested interest in their success, or they build a product that generates its own proprietary data. Or, even better, they find a way to use publicly available data in a novel way. One of my portfolio companies is doing this with mental health, using data from Reddit and other online forums to identify trends and patterns. It’s a brilliant way to get around the data bottleneck.
4. Reimbursement is a Four-Letter Word
In most industries, if you build a great product, people will pay you for it. In healthcare, it’s not that simple. You can have the best product in the world, but if you can’t get it reimbursed by insurance, you don’t have a business. Period.
I’ve seen too many founders treat reimbursement as an afterthought. They build the product, they get it into the hands of doctors, and then they try to figure out how to get paid. That’s backward. You need to be thinking about reimbursement from day one.
Who is going to pay for this? The hospital? The insurance company? The patient? What CPT codes are you going to use? Do you need to apply for a new code? These are not fun questions to think about, but they are essential. The best founders I’ve worked with have a reimbursement expert on their team from the very beginning. They understand that in healthcare, the product is not just the software, it’s the whole package: the software, the clinical validation, and the reimbursement strategy.
5. The Sales Cycle is a Marathon, Not a Sprint
If you’re used to the world of SaaS, where you can close a deal in a matter of weeks, you’re in for a rude awakening in healthcare. The sales cycle for a hospital can be 12, 18, even 24 months. It’s a long, slow, grinding process.
You need to have the cash and the patience to survive it. I’ve seen promising companies die on the vine because they ran out of money before they could close their first big hospital contract. You need to be realistic about how long it’s going to take, and you need to raise enough capital to give yourself the runway you need.
But it’s not just about patience. It’s also about strategy. You need to understand the buying process inside a hospital. Who are the stakeholders? Who is the economic buyer? Who are the champions? Who are the saboteurs? It’s a complex, political landscape. The founders who succeed are the ones who learn how to navigate it.
6. Regulation is Not the Enemy
Most founders, especially in Silicon Valley, have an almost allergic reaction to regulation. They see it as a barrier to innovation, a bunch of red tape designed to slow them down. And they’re not entirely wrong. But in healthcare, regulation can also be a competitive advantage.
If you can get your product through the FDA approval process, you’ve created a massive barrier to entry for your competitors. It’s a long and expensive process, but it’s also a powerful moat. I’ve seen this play out with a company in my portfolio that has an AI-powered device for monitoring a chronic condition. They spent years and millions of dollars on their FDA submission. But now that they have it, they’re the only game in town. No one can touch them.
So don’t be afraid of the FDA. Embrace it. See it as an opportunity to build a defensible business. It’s not for the faint of heart, but if you have the stomach for it, it can be a powerful weapon.
7. It’s Still All About the Team
I know it’s a cliché, but it’s true. At the end of the day, I’m not investing in a model, or a product, or a market. I’m investing in a team. And in AI healthcare, the team is even more important than usual.
You need a team that is not just technically brilliant, but also has a deep understanding of the healthcare industry. You need people who have been in the trenches, who have felt the pain of the problems you’re trying to solve. You need a mix of engineers, doctors, and business people. And you need a founder who is a little bit crazy. Someone who is willing to run through walls to make their vision a reality.
I’ve been lucky to back a few of those founders. And I’m always looking for more. Because as hard as it is, as brutal as the industry can be, I still believe that AI has the potential to transform healthcare. And I want to be a part of making that happen.
Frequently Asked Questions
Do all experts agree with this view?
No, and that's fine. The best ideas in business are often contrarian. I share my perspective based on my experience and data, but I encourage you to seek out opposing viewpoints and form your own conclusions.
How has this view evolved over time?
My thinking on most topics has changed significantly over the years. Early in my career, I held many conventional views that experience proved wrong. I try to update my beliefs when the evidence changes.
What experience informs this perspective?
This perspective comes from over a decade of building companies in Silicon Valley, two successful exits (RemoteTeam to Gusto, MovieLaLa to Gfycat), and investing in 200+ startups including Anthropic, OpenAI, and Scale AI. I write about what I've lived.