Mental healthcare is in crisis. 120M Americans live in areas with clinician shortages. A million attempt suicide every year. Crisis hotlines are underfunded and overburdened. Most people experiencing serious mental health disorders don’t get treatment. Diagnoses are often complicated to formulate and frequently inaccurate. So... how do we get out of this bind? We argue the answer lies in an abundantly available resource: language. We’ll talk about tech that predicts suicide attempts 10X more accurately than a doctor’s evaluation, algorithms that raise the red flag for onset of psychosis, data-driven improvement of unreliable diagnostic categories, and more. And as a technologist/psychologist duo, we’ll cover the ethical and practical challenges of putting these technologies into practice.