I’ve been in the Silicon Valley trenches for over a decade. I’ve built and sold two companies—RemoteTeam to Gusto and MovieLaLa to Gfycat—and I’ve written checks for over 200 startups, including some of the biggest names in AI like Anthropic, OpenAI, and Scale AI. I’ve seen hype cycles come and go. But the buzz around AI in mental health? This one feels different. It feels… dangerous.
Every founder and their dog is pitching an AI therapist app these days. They come to me with slick demos of empathetic-sounding chatbots, talking a big game about “democratizing access” to mental healthcare. They see a massive market and a shortage of human therapists, and they think the solution is an algorithm.
Here’s the uncomfortable truth they’re missing: most of them are dead wrong.
They’re so obsessed with the tech, with building the perfect large language model, that they’re completely ignoring the single most important part of therapy: the human connection. A chatbot can’t replace a therapist. Believing it can isn’t just naive; it’s a disservice to people who are genuinely struggling.
The Seductive, Flawed Promise of an AI Therapist
I get the appeal. Really, I do. An AI therapist is available 24/7, it doesn’t judge you for your darkest thoughts, and it’s a hell of a lot cheaper than a human. For someone wrestling with anxiety or depression, talking to a bot can feel less intimidating than opening up to a person. The market opportunity is obviously enormous.
And look, AI is doing incredible things in other areas of healthcare. I’m an investor in several biotech AI companies that are using machine learning for diagnostics and automation in ways that are genuinely revolutionary. AI can scan a million data points and spot a pattern a human doctor would miss. That’s a perfect use case for the technology.
But mental health isn’t a data problem. It’s a human problem.
What Building a Remote Company Taught Me About Connection
When I started RemoteTeam, we were a fully distributed company from day one. This was long before the pandemic made it trendy. It was a constant struggle to build a real, cohesive culture when we were all just faces on a screen. We made a ton of mistakes.
What we learned, the hard way, was that you can’t automate human connection. The things that made our team work were the messy, inefficient, human things. The regular one-on-ones that went off-topic. The virtual happy hours that were kind of awkward but also kind of fun. The open-door policy that meant my co-founder and I were always available for a call, even if it was just to vent.
We built a culture of trust, not an algorithm for productivity. And that’s a huge reason why Gusto acquired us. That experience is burned into my brain. You can’t fake connection.
So, when the AI therapist pitches started flooding my inbox, I was immediately skeptical. I’ve seen this movie before. Founders fall in love with their tech and forget about the actual problem they’re supposed to be solving. They get lost in the weeds of their model and forget they’re dealing with vulnerable people.
But I’ve also seen a handful of founders who get it. They’re not trying to replace therapists. They’re building tools to augment them. And that’s where I think the real, non-b.s. opportunity lies.
How AI Can Actually Help, Not Hurt
Instead of trying to build a digital Freud, we should be building AI tools that free up human therapists to do what they do best. Here’s what that looks like in practice:
- Killing the Paperwork: Therapists are drowning in administrative tasks. AI can handle the session transcriptions, the note-taking, the billing. This is a simple, unsexy, but incredibly valuable use of AI that gives therapists more time to focus on their patients.
- Providing Data-Driven Insights: AI can analyze session transcripts to spot patterns a therapist might miss. It can track a patient’s progress over time, providing objective data to help inform treatment. This isn’t about replacing the therapist’s judgment; it’s about giving them a better map.
- Personalizing Treatment: AI can help tailor treatment to an individual’s specific needs, recommending interventions or exercises based on their symptoms and progress. It can act as a smart assistant, not a replacement.
- Supporting Patients Between Sessions: A chatbot can be a useful tool for support between appointments. It can help a patient practice cognitive behavioral therapy (CBT) exercises or offer a safe space to vent at 3 AM. But it must be framed as a support tool, an adjunct to human therapy, not the therapy itself.
The Red Flags of a Purely AI Approach
I’m still deeply worried about the startups aiming to replace therapists entirely. It’s a minefield of ethical and practical problems.
- The Empathy Illusion: A chatbot can mimic empathy, but it can’t feel it. It can’t understand the subtle, non-verbal cues of human emotion. That genuine connection, the feeling of being truly seen and understood by another person, is the bedrock of healing. You can’t code that.
- Garbage In, Garbage Out: AI models are trained on data, and that data is full of human biases. An AI therapist trained on a skewed dataset could easily perpetuate and even amplify existing inequalities in mental healthcare. It’s a huge risk.
- Privacy Nightmares: We’re talking about the most sensitive data imaginable. Are all these startups really equipped to handle it with the seriousness it deserves? A single data breach could be catastrophic for patients.
- Who’s Accountable?: When an AI therapist screws up—and it will—who’s to blame? The founder? The engineers? The company? We’re in a legal and ethical gray zone with no clear answers.
My Challenge to Founders
If you’re a founder in this space, I’m not telling you to give up. We need your passion and your intelligence. But I am challenging you to be more thoughtful.
Stop trying to automate the therapist out of a job. Start building tools that empower them. Solve a real problem, not a hypothetical one. And for God’s sake, don’t lose sight of the human element.
The path to success in healthcare AI is brutal. It’s a regulated, complex industry for a reason. A slick UI and a clever model aren’t enough. You need a deep, humble understanding of the problem and the people you’re trying to serve.
I’m still bullish on AI. I’m still writing checks. But in mental health, my money is on the founders who are augmenting our humanity, not trying to render it obsolete.
What’s your take? Am I a jaded cynic, or a necessary voice of caution? Drop your thoughts in the comments.
Frequently Asked Questions
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.
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.
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.