''' I still remember the sting of that rejection. We had built what we thought was a revolutionary AI diagnostic tool. The algorithm was brilliant, the accuracy was off the charts in our lab, and we were sure we were going to change the world. We poured millions into it. Then we met the FDA. It was a bloodbath. They didn’t care about our elegant code or our p-values. They wanted to see a level of rigor and clinical validation we were completely unprepared for. That company is now a footnote in my investment history.
That failure taught me a brutal lesson: in the world of medical AI, the normal Silicon Valley rules don't apply. You can't just "move fast and break things" when people's lives are on the line. Building a great AI model is table stakes. The real challenge, and the real opportunity, is navigating the labyrinth of FDA approval. I’ve seen too many brilliant founders crash and burn because they underestimated this. They think it's a bureaucratic hurdle to be cleared at the end. It's not. It's the entire game.
The Hard Truth About Medical AI
Let's get one thing straight. The FDA is not your enemy, but they are not your friend either. They are the gatekeeper, and their only mandate is to protect public health. They have a healthy skepticism of new technology, and for good reason. I once sat in a meeting where a founder, a genius programmer, tried to explain the intricacies of his neural network to a panel of seasoned clinicians and regulators. He was met with blank stares. They didn’t want a lecture on machine learning; they wanted to know how he could guarantee the model wouldn't misdiagnose a patient and cause real harm.
This is the disconnect that kills so many medical AI startups. We, as tech people, fall in love with the technology. We see the beauty in the data, the elegance of the algorithm. The FDA sees a black box that could potentially kill someone. Your job is to bridge that gap. You have to translate your technical achievements into the language of clinical safety and efficacy. It’s a completely different mindset.
My Playbook for FDA Approval
After that first painful failure, I developed a new playbook for any medical AI company I invest in or advise. It’s not about shortcuts or hacks. It’s about a fundamental shift in how you approach building the company from day one.
De-Risk Early and Often. Don't wait until you have a finished product to engage with the FDA. That’s suicide. I push my founders to request a Pre-Submission meeting as early as possible. This is a formal process where you can introduce your technology and your proposed clinical validation plan to the agency and get their feedback. It’s invaluable. You get a direct line into their thinking and can adjust your strategy before you’ve spent millions on a clinical trial that’s doomed to fail. One of my portfolio companies, a startup building an AI-powered tool for early cancer detection, had three Pre-Submission meetings before they even finalized their product design. It felt slow at the time, but it saved them years and millions in the long run.
Build a Regulatory Moat. Most founders see regulation as a burden. I see it as a competitive advantage. If you can successfully navigate the FDA process, you create a massive barrier to entry for your competitors. It’s your "regulatory moat." This means hiring regulatory experts from the very beginning. Not consultants you bring in at the last minute, but full-time team members who live and breathe this stuff. They should be in every product meeting, every strategy session. They are just as important as your lead engineer.
Clinical Trials Are Your Everything. Your fancy AI model is worthless without robust clinical data to back it up. This is where so many companies stumble. They try to get by with small, poorly designed studies. It won’t work. You need to invest in large, multi-center, prospective clinical trials that are designed to meet the FDA’s stringent standards. I advised a company that was developing an AI for diabetic retinopathy screening. Their initial trial was a mess. The data was noisy, the patient population wasn’t representative, and the results were inconclusive. We had to scrap the whole thing and start over. It was a painful and expensive decision, but it was the right one. Their second trial was rigorously designed, and they sailed through the approval process.
The Right Team is Everything. I can't stress this enough. A medical AI company is not a typical tech startup. You need a multidisciplinary team. Yes, you need brilliant engineers and data scientists. But you also need experienced clinicians who understand the disease area, regulatory experts who know the FDA inside and out, and lawyers who can navigate the complex legal landscape. One of the most successful medical AI companies I’ve invested in was founded by a computer scientist and a radiologist. That combination of technical and clinical expertise was their secret weapon.
Real-World Lessons
I’ve seen this play out time and time again. I invested in a company that had a groundbreaking AI for predicting sepsis in hospitals. The technology was incredible. But the founder was a classic Silicon Valley archetype. He was arrogant, he dismissed the regulators as bureaucrats, and he tried to cut corners on his clinical trial. The FDA put him through hell. It took him four years and three re-submissions to finally get clearance. By then, a competitor who had taken a more collaborative and rigorous approach had already captured the market.
On the other hand, I backed a company building an AI-powered platform for mental health. The founder was a former clinician. She understood the system. She built a strong relationship with the FDA from day one. She invested heavily in a well-designed clinical trial. Her company got FDA clearance in less than a year and is now the market leader.
The difference wasn’t the technology. It was the mindset.
The Future is Regulated
Look, I’m a tech optimist at heart. I believe AI has the potential to transform healthcare in ways we can’t even imagine. But I’m also a realist. The path to that future is through the FDA. There are no shortcuts. You have to embrace the process, you have to respect the rigor, and you have to be willing to play the long game. The founders who understand this are the ones who will build the next generation of iconic healthcare companies. The ones who don’t will end up as another footnote in my investment history. '''))oxiaomi/ask/how-to-navigate-the-fda-approval-process-for.md", text=
Frequently Asked Questions
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.
How can I apply this thinking to my own situation?
Start by identifying the core principle behind the opinion, not the specific example. Then ask yourself: does this principle apply to my context? If yes, test it in a small, low-risk way before going all in.
What's the most common pushback you get on this?
People often push back by citing exceptions or edge cases. And they're usually right that exceptions exist. But building a strategy around exceptions rather than patterns is a losing game for most founders.