I’ve seen a lot of things in my years in Silicon Valley. I’ve built and sold two companies, one to Gusto and another to Gfycat. I’ve written checks for over 200 startups, including some you might have heard of like Anthropic, OpenAI, Scale AI, and Hugging Face. But I’ve never seen anything like the AI gold rush in healthcare. It’s a frenzy. And it’s leading to a lot of wasted money.
A recent report pegged the annual waste from poorly implemented AI in US hospitals at over a billion dollars. I think that number is a wild understatement. From what I’ve seen on the ground, both as an investor and an advisor, the real figure is much, much higher. The problem is that everyone is rushing to do something with AI, without a clear idea of what they’re actually trying to achieve.
They’re buying shiny new toys, listening to slick sales pitches, and making the same mistakes over and over again. I believe most founders think building a great AI model is enough. They're wrong. Here's the uncomfortable truth about what it really takes to succeed in the brutal, regulated world of healthcare AI.
The
Shiny Object" Syndrome
I was talking to a hospital administrator a few months ago. Let's call him John. He was incredibly excited about a new AI-powered diagnostic tool he had just purchased for a seven-figure sum. It promised to detect a rare form of cancer with 99% accuracy. Impressive, right? The problem was, this particular cancer was so rare that his hospital saw maybe one case a year. The expensive, shiny new AI was going to sit in a corner, collecting dust.
This is what I call the "Shiny Object" Syndrome. Hospitals are so afraid of being left behind that they're throwing money at any AI solution that looks impressive, without asking the most basic question: what problem are we actually trying to solve? They get mesmerized by the tech and forget about the practicalities. It's a classic case of a solution in search of a problem.
Here’s a hard truth: a 99% accurate model for a problem you don't have is worthless. It's just a very expensive line item on your budget. Before you even think about buying an AI tool, you need to have a crystal-clear understanding of your own pain points. Are you trying to reduce patient wait times? Improve diagnostic accuracy for a common disease? Streamline your billing process? Start with the problem, not the solution.
Ignoring the Workflow
Let's say you've identified a real problem. You've found an AI tool that can solve it. You're golden, right? Not so fast. The biggest hurdle to AI adoption in hospitals isn't the technology itself. It's the workflow.
Hospitals are complex, chaotic systems. Doctors and nurses have been trained to do things a certain way for years. Their workflows are ingrained, and for good reason. They're designed to minimize errors and save lives. If your fancy new AI tool doesn't fit seamlessly into their existing workflow, they're not going to use it. It's that simple.
I once invested in a startup that had developed a brilliant AI for radiologists. It could analyze medical images with incredible speed and accuracy. But the interface was clunky. It required the radiologist to open a separate application, upload the image, and wait for the results. It was a great piece of tech, but it was a terrible product. It added friction to the radiologist's day, instead of removing it. The company eventually went under. It was a painful lesson.
The best medical AI tools are invisible. They work in the background, integrated directly into the software that clinicians are already using. They don't require extra steps or new logins. They just provide the right information, at the right time, in the right place. That's the holy grail.
Garbage In, Garbage Out
Everyone in the AI world knows the phrase "garbage in, garbage out." It means that the quality of your AI model is only as good as the data you train it on. But in healthcare, this isn't just a technical issue. It's a matter of life and death.
Hospital data is a mess. It's fragmented, it's inconsistent, and it's often just plain wrong. You have data from different departments, in different formats, with different standards. Trying to train an AI model on this kind of data is like trying to build a house on a foundation of sand. It's not going to work.
I've seen companies spend millions of dollars and years of their lives trying to clean up messy hospital data. It's a thankless, brutal task. And even then, you can never be sure that you've caught all the errors. A single mislabeled data point could lead to a catastrophic misdiagnosis.
This is why I'm a huge believer in a new approach: what I call "workflow-first" data collection. Instead of trying to clean up messy data after the fact, you design your workflow to generate clean, structured data from the very beginning. You build data validation and quality checks directly into the tools that clinicians are using every day. It's more work upfront, but it pays off a thousand times over in the long run.
The Human Element
Here's the most important lesson I've learned about AI in healthcare: it's not about replacing doctors. It's about augmenting them. It's about giving them superpowers.
The best AI tools don't try to be the hero. They're the trusty sidekick. They handle the tedious, repetitive tasks that burn doctors out, so they can focus on what they do best: thinking critically, connecting with patients, and making tough judgment calls. They're a second pair of eyes, a safety net, a cognitive assistant.
I remember talking to a radiologist who was using an AI tool to help him screen for lung cancer. He told me, "This thing doesn't make me a worse radiologist. It makes me a better one. It catches things I might have missed at the end of a long day. It gives me more confidence in my decisions." That's the sweet spot. That's where the magic happens.
So, what's the takeaway here? It's simple. Stop chasing shiny objects. Start with your problems. Obsess over your workflow. Get your data house in order. And never, ever forget the human element. That's how you build healthcare AI that actually works. That's how you stop wasting billions of dollars and start saving lives.
Frequently Asked Questions
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.
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.
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.