The mental health crisis isn't about medication. It's about bandwidth.

If you listen to the public conversation about mental health in America, you'd think we had a medication problem. We don't. We have a bandwidth problem.

Here are the numbers.

There is one psychiatrist for every 1,200 Americans with a mental health condition. Half of U.S. counties have zero psychiatric coverage at all. The average wait for a first appointment is more than three months. And despite all of that, we somehow manage to write a staggering volume of psychiatric prescriptions — 80% of SSRIs in this country are prescribed by someone who isn't a psychiatrist. A family doctor with seven minutes. An urgent care NP. A telehealth refill mill.

Access to drugs is not the bottleneck. Access to real, continuous, thoughtful psychiatric care is.

What "real care" looks like — and why the system can't deliver it

A good psychiatric relationship is not a prescription. It's a longitudinal conversation. The psychiatrist listens for what changed since last week. Sleep. Appetite. The quality of your thinking. The side effect you almost didn't mention. They titrate. They adjust. They notice the spiral two weeks before you do.

That is what works. It's also what the traditional system is structurally incapable of delivering at scale. A 15-minute appointment every three months is not continuity. It's a refill ritual dressed up as care. And the outcomes show it: about 60% of patients drop off treatment within a year, about 60% relapse within 12 months, and hospitalizations are climbing.

Where AI actually helps

This is the part most AI-for-mental-health stories get wrong. The opportunity is not "AI therapy" — a chatbot that role-plays a counselor. There are 300 of those. The opportunity is the unglamorous part of psychiatry: the daily listening, the pattern recognition, the adherence nudges, the early-warning signals. The parts no human can do at the scale this country needs.

Done right, AI does three things a human psychiatrist cannot:

• It's always on. Symptoms don't respect office hours. AI does its best work in the gap between appointments — which is where care actually happens.
• It remembers everything. Every mood, every side effect, every conversation. You never retell your story.
• It sees patterns across weeks. The signal that says "something is shifting" shows up long before the crisis. A quarterly appointment can't see it. Daily check-ins can.

The catch: none of this is safe without a clinician in the loop. Diagnoses have to be reviewed. Prescriptions have to be signed. Crises have to escalate to humans. That's the model we're building toward — AI-native, never AI-only.

The shift that's already underway

More than 50 million Americans already use AI tools for mental health — ChatGPT, symptom checkers, journaling apps. For a specific task (diagnosing OCD), AI is now roughly twice as accurate as human clinicians in head-to-head studies. Utah just greenlit AI for routine prescription refills. The Health Technology Act, introduced to Congress last year, would formally allow AI to prescribe under clinician oversight.

The regulatory, clinical, and consumer tailwinds are all pointing the same direction. The question isn't whether AI becomes the foundation of mental healthcare. It's who builds it responsibly, and who cuts corners.

What we're doing about it

We started Siggy because we'd spent twenty years building healthcare companies that bumped into the same wall: you can make care cheaper, faster, and easier to access — but until you solve the bandwidth problem at the clinician layer, you're just rearranging the furniture.

A full-stack AI psychiatrist, clinically supervised, available every day, priced so anyone can afford it: that's the wall coming down.

Mental wellness should be the norm, not the exception. It's going to take a new kind of infrastructure to get there. We're building it.

— The Siggy team