Jimini Health

Jimini Health builds Sage, a clinician-supervised, patient-facing AI behavioral health assistant that works as a member of the care team rather than a standalone chatbot. Sage supports patients 24/7 before, between, and after therapy sessions (coping-skills practice, mood tracking, journaling, structured exercises, medication-adherence support, proactive check-ins), gathers intake data, tracks measurement-based outcomes (PHQ-9/GAD-7), surfaces risk signals with escalation pathways, and keeps every interaction visible to licensed clinicians who set its priorities. The company runs its own multi-state therapy practice as a proving ground and is now selling Sage as clinical-grade patient-facing AI infrastructure to large behavioral health provider organizations, with EHR integration and compliance tooling.

Stated: keep licensed clinicians in control while making patient-facing LLMs 'a core part of patient care in behavioral health' — sell Sage as compliant, supervised AI infrastructure to the largest behavioral health provider organizations, expand across comorbidities, care settings, and engagement modalities, and publish safety frameworks to shape industry norms (CEO in June 2026: organizations that move early on responsible clinician-supervised models will set the standards everyone else follows). Inferred: a staged-autonomy land-and-expand play — prove safety and outcomes inside its own clinic, use that data plus safety branding to win enterprise behavioral health systems via a Palantir-style forward-deployed-engineer motion (reflecting the founders' Palantir DNA), then ride outcome-based reimbursement (CMS ACCESS model, Medicare Advantage/Medicaid via Town Hall Ventures' network) toward progressively more autonomous AI care as trust and regulation permit; effectively positioning as the regulated B2B counterweight to consumer AI-therapy apps.

Thin but forming: proprietary longitudinal data from supervised AI-patient interactions in its own clinic, purpose-built safety/supervision infrastructure (suicidality/psychosis risk classifiers, clinician oversight consoles, EHR and compliance integration) that consumer chatbots lack, in-house model work (the Head of AI role specified supervised fine-tuning, RLHF, and red-teaming rather than thin wrappers on frontier APIs), and reputational capital from an elite clinical-academic advisory network (DeepMind, Harvard, Yale, Stanford, MIT) plus published safety frameworks. No structural lock-in yet — the moat depends on becoming the trusted default for enterprise behavioral health before better-funded players do.