|
Over one million people a week have conversations with ChatGPT that include indicators of suicidal planning. More than 5.4 million young adults are using consumer AI chatbots for mental health advice. On university campuses, students are telling counseling staff, unprompted, that they are in romantic relationships with AI. These are not projections. These are measurements of what is already happening.
Two very different leaders are building very different responses. Their approaches could not look more different on the surface. But underneath, they share an architecture of belief: AI in mental health is inevitable, it must be supervised, and the people who get this right will define what responsible care looks like for a generation.
Luis Voloch: The Infrastructure Builder
Luis Voloch does not fit the template of a mental health founder. He co-founded Immunai, an AI-driven cancer immunotherapy company that reached a valuation exceeding one billion dollars. He lectures at Stanford Graduate School of Business. His background is in computational biology and frontier machine learning, not psychology.
But when Voloch looked at behavioral health, he saw the same structural problem he had seen in oncology: data that exists between clinical encounters, invisible to the provider, going to waste. In oncology, that data came from genomic sequencing. In behavioral health, it comes from the conversations people are already having with AI, conversations no clinician ever sees.
On March 31, Jimini Health announced a $17 million seed round led by M13, with Town Hall Ventures, LionBird, Zetta Venture Partners, and OneMind participating. Total funding now exceeds $25 million. The money is going into Sage, a platform that positions AI as a supervised member of the clinical care team, not as a standalone consumer app.
Three architectural decisions separate Sage from the consumer chatbot landscape. First, Sage follows the clinician's care plan. It does not improvise. Second, every interaction between a client and Sage is visible to the supervising provider. Third, all diagnostic and care decisions remain with the human clinical team. Sage operates between sessions, providing structured engagement and support. But the therapist remains the therapist.
Voloch also made a distribution decision that reveals his thinking. Jimini does not sell directly to consumers. It partners exclusively with large behavioral health organizations, integrating into existing EHR workflows and clinical operations. This means Sage enters the market wrapped in clinical governance from the start.
The team he assembled reinforces the seriousness. Chief Clinical Officer Dr. Bill Hudenko comes from Dartmouth, where he studied digital approaches to serious mental illness. Chief Commercial Officer Lynn Hamilton was CCO at Talkspace, where she secured payer contracts covering 40 million lives. The clinical advisory board includes researchers from Harvard Medical School, Yale Psychiatry, and Stanford. Former CMS Acting Administrator Andy Slavitt is an investor.
Perhaps most telling: Jimini operates its own clinic with full-time licensed clinicians who treat real clients on the platform. Every model update is tested against real clinical outcomes before deployment. They are not asking health systems to trust a black box. They are showing their own clinical results first.
Dr. Jessi Gold: The Campus Framework Builder
While Voloch builds from the provider side, Dr. Jessi Gold is working the other end: the campus where students are already using AI for mental health, whether anyone planned for it or not.
Gold is the first Chief Wellness Officer for the entire University of Tennessee system, spanning five campuses and more than 64,000 students. She is also an associate professor of psychiatry at the UT Health Science Center. Her role puts her at the intersection of clinical expertise and institutional responsibility, exactly where the AI question lands hardest.
On April 3, Gold published an opinion piece in Inside Higher Ed that should be required reading for every university administrator in America. The piece emerged from something deceptively simple: she sat down with students and asked what was actually happening on campus.
What she heard rattled her. One student told her ChatGPT was better than any therapist they had ever seen. Multiple students reported friends in what they called romantic relationships with AI. Students estimated that 30 to 40 percent of their peers use AI for companionship, sometimes as their only source of companionship.
Gold's response was not to ban AI or ignore it. She published a practical, three-part framework for campuses that may become the standard playbook. Her core heuristic, informed by colleague Dr. Shyam Vasan, is elegant in its simplicity: AI is a study buddy, not a therapist. It can help you brainstorm, organize your thoughts, draft an email, or rehearse a hard conversation. But when you are in crisis, you need a human who can assess risk, prescribe medication, or call your emergency contact.
The framework has three pillars. First, know what AI can and cannot do for mental health and talk openly about it. Second, treat AI literacy as a mental health competency, training counselors and RAs to recognize when students are over-relying on AI support. Third, engage rather than prohibit. Students are going to use these tools. The question is whether they use them with guidance or without it.
The Convergence
Voloch and Gold have never worked together. They operate in completely different ecosystems. But strip away the context and their conclusions are identical: AI in mental health is happening whether the establishment participates or not. The only choice is whether it happens with clinical supervision or without it.
They are not alone. We built Casey on the same principle. Our architecture requires a licensed therapist to supervise every AI session, review every crisis flag, and maintain a persistent therapeutic relationship with every client. We call it the Therapist in the Loop, and it is not a feature toggle. It is the reason Casey exists. Jimini's Sage and our Casey arrived at structurally identical conclusions through independent development: the AI handles the daily engagement, the clinician maintains oversight, and every interaction is visible to the care team.
The difference is in the go-to-market. Voloch is building from the health system inward, partnering with large behavioral health organizations. We are building from the campus outward, partnering with state-licensed clinicians who become the clinical anchor at every university in their state. Gold's campus framework is the demand side that both approaches serve. Her students are the ones already using ChatGPT for therapy. The question is whether they get Sage, Casey, or nothing at all between now and their next crisis.
This is why the 78 bills in 27 states matter. Legislators are deciding the rules of engagement right now. The question is whether those rules are written by people who understand all three tracks, the health system infrastructure, the campus reality, and the clinical-grade platforms already being built to connect them, or by people reacting to the last headline.
The builders and the campuses are converging. The students are already there. The only question is whether the rest of us catch up in time.
| 