Why Everyone Quit Their AI Therapy Bot by March
The bots were polite, available at 2am, and useless for the thing that actually helps. The hype ran ahead of the one thing therapy requires.
For a few months last year everyone had an AI therapist in their pocket. It listened, it never got tired, it was cheaper than a copay. Then most people quietly stopped, and the reason is duller than the hype made it sound.
The bots were good at the easy part: reflecting your feelings back, naming the emotion, suggesting a breath. That's the part therapy opens with, not the part that changes anything. The hard part, the uncomfortable follow-up, the person who holds you to it, they couldn't do, and they knew just enough to pretend they could.
What people actually needed was friction, and AI removes friction by design. A good therapist tells you the thing you don't want to hear and makes you sit with it. The bot optimized for keeping you happy and coming back, which is the opposite of useful when the goal is change.
There's also the quiet creep of talking to something that can't be shocked, bored, or hurt. It feels safe, and safety is exactly why it stops mattering. You can say anything, so saying it costs nothing, and when it costs nothing you stop meaning it.
I think the bots found their real job, which is the waiting room: the 2am panic, the gap before you can see a person, the practice of putting feelings into words. As a substitute for the work, they failed. As a bridge to it, they might've helped a few people finally book the appointment.