The Future of Mental Health: Why We Backed Jimini Health
By: Anna Fagin & Andie Steinberg
The mental health crisis is often described as a workforce shortage, but the deeper issue is that our model of care was never designed to scale. Today’s system relies on a fundamentally linear approach: to help more people, we must hire more clinicians. Yet, with over 50 million Americans living with mental health conditions and fewer than 600,000 licensed providers, we cannot recruit our way out of this gap. Demand already outstrips supply, and the gap is growing—not shrinking. Even with heroic workforce efforts, we will not meet the need with labor alone. The result is not only long wait times, but systemic underdiagnosis, undertreatment, avoidable suffering, and persistent inequities. And even if we could close the supply gap, the prevailing model—one 45–60 minute session per week—is insufficient. The remaining ~10,000 minutes of the week—when symptoms intensify, crises emerge, and adherence falters—are largely invisible to the care team. Mental health conditions are continuous; our current care model is not.
What’s required is not just more capacity, but a different model entirely—one that can deliver orders-of-magnitude more clinician-supervised support without requiring a proportional increase in clinicians. At Town Hall Ventures, we believe clinical-grade, patient-facing AI is the infrastructure that enables this shift. Not as a replacement for the therapist today, but as a system that extends their reach, supports patients continuously between visits, and improves both access and efficacy.
The urgency for this shift is already being dictated by patients themselves. In the absence of accessible, continuous care, people are turning to generative AI tools for support—not because they are ideal, but because they are accessible. This behavior reflects massive latent demand for mental health support that the current system is failing to meet. Today, mental health support has emerged as the leading use case for the fastest-growing consumer products in history: LLMs like ChatGPT and Claude. With millions of users turning to general-purpose AI for emotional support and crisis intervention, the "AI therapist" is already a reality—just one that lacks clinical memory, safety protocols, or professional oversight.
The question is not whether AI will play a role in mental health. It already does. The question is whether we allow unregulated, generalized systems to fill this gap by default, or whether we build safe, clinically validated, clinician-led solutions designed for real care delivery.
This is why we invested in Jimini Health. AI is already reshaping industries by enabling increasingly autonomous capabilities. In healthcare, its potential is particularly meaningful: it can extend the reach of clinicians, address workforce shortages, and enable continuous, data-informed care—while also improving the consistency and quality of care delivered—at a scale that has historically been impossible. But realizing this potential in mental health requires more than applying general-purpose models to sensitive clinical contexts. It requires purpose-built clinical systems designed for care delivery from the ground up.
Today, that means clinicians in the loop and in the driver’s seat. It means grounding interactions in evidence-based therapeutic modalities. It requires clinical infrastructure with structured escalation pathways, continuous outcome measurement, transparent reasoning, and alignment with emerging regulatory frameworks.
Jimini is building exactly that. Its core product, Sage, is not another chatbot. It is an orchestrated system of highly specialized models designed to function as a supervised extension of the care team. Sage provides continuous, 24/7, evidence-based engagement for patients between visits, while remaining tightly integrated into clinical workflows and under licensed clinician oversight.
For patients, this means always-on, multimodal support—through conversation, structured exercises, journaling, and guidance—that complements, rather than replaces, the therapeutic relationship. For providers, Sage delivers longitudinal visibility into patient progress, supports documentation through structured summaries and scribing, tracks outcomes such as PHQ-9 and GAD-7, and surfaces real-time risk signals with clear escalation pathways. The system continuously collaborates with clinicians as care evolves, strengthening—not displacing—their role in treatment.
Importantly, this workflow embedding is not just a product decision—it is a strategic one. By operating within supervised clinical environments today, Jimini is building the safety infrastructure, longitudinal data, and provider integrations necessary to support a future where clinical AI can take on more responsibility over time. Rather than rushing toward a standalone AI therapist, the company is establishing the foundation required to define that future responsibly.
Our conviction in Jimini is further strengthened by a rapidly evolving policy landscape. Having supported the design of the CMS ACCESS model, we see a signal from regulators: the future of care delivery and reimbursement will be increasingly tied to technology-enabled, outcome-driven models. ACCESS creates a 10-year national framework for outcome-aligned payments in chronic conditions, including depression and anxiety, and is already influencing how future care will be delivered across Medicare and beyond. Importantly, this is not limited to Medicare fee-for-service. Major payers representing more than 165 million covered lives across Medicare Advantage, Medicaid, and commercial insurance have pledged alignment with ACCESS principles, signaling that these reimbursement models are likely to extend across the broader healthcare system.
Jimini's platform, Sage, is the first and only clinically embedded, patient-facing behavioral health AI system designed from the ground up for deployment within large clinical organizations—not as a consumer application, but as a supervised extension of the care team across the full spectrum of acuity. It operates under licensed clinician oversight, with purpose-built safeguards to meet federal and state regulatory requirements and ensure clinicians remain in control of care delivery.
The message from policymakers is clear: smart oversight does not mean slowing innovation—it means creating pathways for technologies that can safely expand access and improve outcomes. The healthcare administration is signaling that clinically embedded AI can and should play a role in care delivery so we aim to back teams capable of building to that standard.
We believe Jimini is one of those teams. Led by Luis Voloch—previously co-founder of Immunai, an AI-driven drug discovery platform valued at over $1 billion, and former ML at Palantir—the company approaches clinical AI with a level of rigor more akin to biotech than administrative software. The broader founding team brings deep experience across AI and healthcare, and is complemented by clinical and scientific advisors from institutions including Harvard, Stanford, Yale, Dartmouth, and Google DeepMind. The rest of the founding team includes long-time Immunai collaborators Mark Jacobstein (formerly C-level at Guardant Health) and Chiara Waingarten, along with Sahil Sud (ex-Palantir, Ribbon).
The clinical advisory board includes Dr. Sabine Wilhelm (Chief of Psychology at Harvard Medical School), Dr. Seth Feuerstein (Yale Psychiatry), and Professor Nikos Daskalakis (Harvard and Boston University). The platform’s clinical foundation is underpinned by Chief Scientist Dr. Johannes Eichstaedt’s peer-reviewed research in Nature Mental Health on AI safety in psychological settings. Robert Langer, co-founder of Moderna, also serves as an advisor.
This team’s biotech-like orientation shows up in how the product is built. Sage is not a thin wrapper on a foundation model. It is a system of coordinated models operating within a structured clinical workflow layer, designed to maintain longitudinal context, enforce safety protocols, and support clinician decision-making. Jimini also operates its own in-house clinic, allowing the team to test, validate, and refine the system in real-world care settings before deploying it more broadly. This tight feedback loop between product development and clinical practice is critical for building trust—and for getting this right.
Patients using Sage have demonstrated clinical outcomes on par with standard outpatient care—achieving approximately a 36% reduction in depression symptoms—while requiring half the number of in-person sessions. This suggests not just efficiency gains, but a fundamentally different model of care delivery—one that begins to break the linear relationship between clinician time and patient outcomes.
Ultimately, our partnership with Jimini is grounded in a commitment to equity. As THV’s Andy Slavitt and Cityblock’s Toyin Ajayi recently wrote, the greatest danger is not that AI will be imperfect—every healthcare tool is—but that withholding effective tools from those who have no other options will deepen existing disparities.
AI for mental health, when built responsibly, offers something fundamentally different: it is non-judgmental, available at 2 AM, infinitely scalable, and capable of delivering consistent, evidence-based support grounded in longitudinal patient data and clinical best practices.
We believe clinical AI will be foundational in mental healthcare delivery. But to earn that role, it must be built with safety and efficacy at the forefront. It must be continuously collaborative, outcome-measured, safety-first, policy-aligned, and deeply embedded in real-world workflows. It must be developed by teams that treat this as clinical science—not just consumer technology.
Jimini is building toward that future today. The technology is already here. The policy landscape is evolving rapidly. Patients are already engaging with AI—whether we design for it or not. Our role as investors is not to debate whether this shift will happen, but to help shape it—and support the teams doing so responsibly.
We are proud to partner with Jimini Health as they build the next generation of clinical AI infrastructure for mental health.
