'''
The AI Model Is the Easy Part. Seriously.
Everyone in Silicon Valley is chasing the next big thing in AI. I get it. I’ve backed over 200 companies, including some of the names you see in the headlines every day like Anthropic and OpenAI. But I’m going to tell you something that might surprise you, especially when it comes to the AI gold rush in healthcare.
Building a great AI model is the easy part.
I’ve seen brilliant teams with PhDs from Stanford and MIT build breathtakingly accurate diagnostic models. Models that could spot cancer on a scan better than a human radiologist. And I’ve seen those same companies die a slow, painful death. Why? Because in healthcare, the tech is just the ticket to the game. It’s not the game itself.
I’ve been in the trenches of this world for over a decade, first as a founder with two exits under my belt, and now as an investor. I’ve put my own money into more than 20 AI healthcare startups, and I’ve got the scars to prove it. You can’t just slap a .ai domain on a problem and expect to win. The world of healthcare is a brutal, regulated, and deeply human system. If you don’t understand that, your beautiful algorithm is worthless.
Here are the seven most important things I’ve learned from the front lines.
1. “Move Fast and Break Things” Will Get You Sued
The classic Silicon Valley mantra is poison in healthcare. When a “broken thing” is a misdiagnosis or a privacy breach, you’re not just looking at a bad press cycle; you’re looking at lawsuits and FDA audits. One of the first companies I invested in had a brilliant idea for automating patient intake. The team was sharp, they built a slick app, but they used a third-party transcription service that wasn’t HIPAA compliant. The mistake cost them a pilot with a major hospital system and nearly killed the company. In healthcare, the rules aren’t suggestions. They are the unforgiving laws of the land.
2. Your Algorithm is Starving for Good Data
Every AI founder knows that data is the fuel. What they don’t realize is that healthcare data is a nightmare. It’s siloed in a dozen different EHR systems that don’t talk to each other. It’s messy, unstructured, and full of human error. One of my portfolio companies in the AI diagnostics space spent the first two years of its life just trying to get a clean, labeled dataset from a single hospital. They had a world-class model, but it was sitting idle, starving for data. The most successful companies I’ve seen are the ones that solve the data problem first, often by providing a tool that clinicians want to use, which generates the proprietary data they need as a byproduct.
3. Doctors Are Your Toughest Customer
If you think you can just walk into a hospital and tell a surgeon who’s been practicing for 20 years that your algorithm is better than their intuition, good luck. Doctors are overworked, skeptical of new tech, and fiercely protective of their patients. Your product can’t just be 10% better. It needs to be 10x better, and it has to seamlessly fit into their existing workflow. It can’t add more clicks. It can’t require a new login. It has to save them time, reduce their administrative burden, or give them a superpower they didn’t have before. The best pitches I’ve seen weren’t about the AI; they were about solving the doctor’s most annoying problem.
4. The AI Radiology Gold Rush is Over
Five years ago, every other pitch I heard was "AI for reading X-rays." It was the obvious first target, the low-hanging fruit. Today, that market is incredibly crowded. There are hundreds of companies with FDA-cleared algorithms for detecting nodules on a chest CT. It’s a race to the bottom on price. The real opportunity isn’t in the sexy, high-profile diagnostic fields anymore. It’s in the boring, unglamorous plumbing of the healthcare system. Think AI for automating insurance pre-authorizations. AI for optimizing hospital bed allocation. AI for reducing patient no-shows. That’s where you can build a massive, defensible business.
5. It’s a Feature, Not a Product
This is probably the biggest mistake I see first-time founders make. They fall in love with their technology and try to sell "AI." Nobody buys AI. They buy solutions to their problems. One of my most successful investments in this space is a company that helps mental health providers with their billing. Their secret sauce is an AI that automatically codes therapy sessions based on clinical notes. But they don’t sell an "AI Coder." They sell "Get Paid Faster and More Accurately." The AI is a feature, a powerful one, but it’s in service of a clear business outcome. The technology is the how, not the what.
6. The Sales Cycle is a Marathon, Not a Sprint
In the SaaS world, you can close a deal in a week. In healthcare, you’re lucky if you can close a deal in a year. You’ll do a pilot. Then a second pilot. Then a security review. Then a legal review. Then a purchasing committee review. Each step can take months. You need a war chest to survive this process. I’ve seen too many startups run out of cash while waiting for a hospital to make a decision. You have to have the capital and the patience to withstand a sales cycle that would make any other tech founder cry.
7. Your Exit Will Be a Strategic, Not a Unicorn
While everyone dreams of a multi-billion dollar IPO, the reality in AI healthcare is that the most likely exit is a strategic acquisition by a large incumbent. A Medtronic, a Philips, a UnitedHealth Group. These giants are not tech companies. They are slow-moving behemoths that are desperate for innovation. They will pay a premium for a company that has a validated product, a foot in the door with hospital systems, and a clear ROI. The exit might not be as fast or as flashy as a consumer tech company, but it can be just as lucrative if you build something of real, durable value.
Investing in AI healthcare isn’t for the faint of heart. It’s a long, hard road. But the problems are some of the most important in the world. If you can navigate this maze, you’re not just building a company; you’re fixing a system that impacts every single one of us. And that’s a mission worth fighting for. '''
Frequently Asked Questions
How were these items selected?
Each item on this list comes from direct experience, either from building my own companies or from patterns I've observed across the 200+ startups I've invested in. I prioritize practical, actionable items over theoretical concepts.
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.
Are these recommendations still relevant in 2026?
Absolutely. While specific tools and tactics change, the underlying principles remain consistent. I update my thinking regularly based on what I'm seeing in the market and across my portfolio companies.