Founders who close a big round for an AI mental health app all say the same thing: it was one of the most brutal fundraising processes they’ve been through. I believe them. I’ve seen two of my companies get acquired and have personally invested in 234 startups, including some of the biggest names in AI like Anthropic and OpenAI.
Everyone is throwing money at AI right now, especially in healthcare. You’d think it would be easy. It’s not. Most founders I talk to are dangerously naive about what it takes to build and fund a company in this space. They think a clever model and a slick pitch deck are enough. They’re dead wrong.
Here’s the uncomfortable truth about what it really takes to succeed in the brutal, regulated world of healthcare AI.
The Idea Is The Easy Part
The idea for a company like this usually comes from a deeply personal place: watching friends and family struggle to find good mental healthcare. Therapists are overbooked, costs are skyrocketing, and the quality of care is inconsistent. It’s a system at its breaking point. AI can help, not by replacing therapists, but by augmenting them, making them more effective and accessible.
Picture a platform that uses AI to analyze therapy sessions (with full consent, of course) and provide actionable insights to both the therapist and the patient. It can identify patterns in speech, detect cognitive distortions, and track progress over time. It’s a tool to superpower therapists, not render them obsolete.
That’s the vision. It gets a team in the door with investors. But it isn’t what gets them the money.
Welcome to the FDA, a.k.a. "The Great Filter"
The moment you say “healthcare AI,” you’re playing in a different league. You’re not just building an app; you’re building a medical device. And that means you have to deal with the FDA. For most tech founders, this is a foreign and terrifying concept.
I remember one of my early angel investments, a brilliant team out of Stanford working on an AI-powered radiology tool. They had a model that could detect certain types of cancer with 99% accuracy in a lab setting. They thought they were going to change the world overnight. They burned through their seed money in 18 months just trying to figure out the paperwork for their 510(k) clearance. The company died before it ever saw a single real patient.
That lesson was seared into my brain. The mental health teams that get it right hire regulatory experts before they write a single line of production code. They spend their first months, and a real share of their seed round, just on mapping out their regulatory strategy. They treat the FDA approval process not as a hurdle, but as a core feature of the product.
Investors see this, and it immediately sets a team apart. They aren’t just tech people with a cool algorithm; they’re a serious company that understands the battlefield. They can show a clear, albeit long and expensive, path to market. That’s what investors invest in.
Your Model Is a Commodity. Your Data Is The Moat.
Another hard truth: your AI model is not that special. With the rise of open-source models and platforms like Hugging Face (one of my portfolio companies), the barrier to entry for building sophisticated AI has collapsed. If you could build it, a dozen other teams can too.
What they can’t replicate is your proprietary data. In healthcare, data is everything. But getting it is a nightmare. You have to navigate HIPAA, build trust with hospitals and clinics, and ensure patient privacy is absolutely sacred. It’s a slow, painful, relationship-driven process.
In a fundraise like this, the sharpest VCs, partners at top-tier firms, can spend almost the entire meeting grilling the founder on data acquisition strategy. Not a single question about model architecture. They ask about partnerships with therapy clinics. They ask about the data anonymization process. They ask about the consent flow for patients.
They know that the company that wins is the company that has the best data. A team that can show signed partnership agreements with a network of therapists in its beta program, an end-to-end encrypted data pipeline, and a detailed document on its HIPAA compliance and data governance policies has a moat. The AI model is just the machine that runs on top of it.
The $25 Million Grind
Even with a solid regulatory strategy and a data moat, the fundraising process is a grind. Expect to pitch a lot of investors. Expect most of them to say no.
The “no’s” are brutal. Some don’t believe in the market. Some think it’s too crowded. Some are scared off by the regulatory risk, even with a detailed plan. One line founders in this space hear: “I love the idea, but I don’t invest in things that could get people sued.”
But the “yes’s” come from the investors who understand the deep, systemic problems in healthcare. They’re the ones who have seen the failures of purely tech-focused approaches. They appreciate an obsession with the unsexy, grinding work of compliance, data acquisition, and building trust with clinicians.
In the final meeting with a lead investor, the conversation that seals the deal often isn’t about the tech. It’s about a story. Picture a therapist in the beta program who is on the verge of burnout. She’s managing 30 patients and feels like she’s failing them. The tool helps her spot a critical pattern in an at-risk patient that she had missed. It doesn’t only help that patient; it restores her faith in her own ability to be a great therapist.
That’s what the best teams are building. Not just an AI company, but a company that understands the human element of healthcare. They’re not just shipping code; they’re trying to solve a real, painful problem for millions of people.
Raising $25 million in this space is incredibly difficult. When it happens, it validates that approach. The hard-won lessons from my past startups and investments all point to this: in healthcare AI, you win by focusing on the boring stuff. The regulations. The data. The trust. The real-world workflow of the professionals you’re trying to help.
Most founders want to build a great AI model. That’s the fun part. But it’s not enough. If you want to succeed in this world, you have to be willing to do the hard, unglamorous work that others won’t. That’s the only way to build something that lasts.
Frequently Asked Questions
Q: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.
Q: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.
Q: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.