# Jimini: clinician-led AI support between behavioral-health sessions

Canonical: https://mudpie.ai/companies/jimini/
Breadcrumb: [Home](https://mudpie.ai/) / [Companies](https://mudpie.ai/companies/) / [Jimini: clinician-led AI support between behavioral-health sessions](https://mudpie.ai/companies/jimini/)
Author: Ali Abouelatta (https://mudpie.ai/authors/ali-abouelatta/)
Published: 2026-09-19
Updated: 2026-09-19
Research type: Company profile
Method: Company and accelerator sources checked 2026-09-19. Product claims are attributed to their sources; this is research, not a hands-on product trial.

Jimini is building a behavioral-health care team where an AI assistant supports licensed clinicians between sessions. The product is aimed at providers and care organizations that want continuity, reminders and skills practice without pretending a chatbot can replace a clinician.

## What it does

The [current Jimini site](https://jiminihealth.com/) describes Sage as an AI care-team member that supports patients before, after and between therapy sessions. It says Sage follows the clinical team’s instructions, helps patients practice skills, shares progress with providers and escalates necessary situations. Jimini explicitly says Sage is never a substitute for human clinicians and presents the product as clinically integrated and clinician overseen.

The company says Sage is built by psychologists, psychiatrists and AI experts and uses evidence-based practices and measurement-based care. Its public site also claims the system is secure and privacy protected, but it does not publish the complete regulatory, security or clinical-outcomes detail a health system would need before deployment.

## Why I’d look closer

The advantage is continuity. A patient’s care plan does not disappear between appointments, and a clinician can see progress rather than rely on memory at the next session. The founder context is unusually relevant: Luis Voloch is described as a Stanford GSB AI researcher and lecturer and co-founder of Immunai; Mark Jacobstein has led healthcare and technology companies including Guardant Health and Immunai.

The tradeoff is safety and scope. Behavioral-health conversations can change quickly, and escalation must work when a patient is distressed, ambiguous or outside the product’s intended use. “Always accessible” is useful only when the care team can define boundaries, response times, crisis paths and data access.

## What I’d ask

Which patient populations and conditions are supported? How does Sage detect and escalate crisis language, self-harm risk or clinical deterioration? Can clinicians inspect the conversation, instructions and model changes? What consent, retention, encryption and audit controls apply? Which claims are clinically validated, and which remain product direction?

## My editorial take

Shortlist Jimini for a clinician-led behavioral-health pilot where between-session support is the measurable gap. Keep the human care team in control, define escalation and crisis procedures before onboarding patients, and evaluate adherence and outcomes through the provider—not through an AI engagement number alone. This is clinical infrastructure, not a self-serve therapy replacement.

## Quick facts

| Field | Sourced detail |
| --- | --- |
| Product | AI behavioral-health assistant integrated with licensed clinicians |
| Buyer | Behavioral-health providers and care organizations |
| Core controls claimed | Clinician instructions, escalation and provider progress sharing |
| Pricing/status | Pricing and complete regulatory status not published in checked pages |
| Main question | Can the assistant extend care safely without blurring the clinician boundary? |

## Sources checked

| Source | Checked |
| --- | --- |
| [South Park Commons Jimini profile](https://www.southparkcommons.com/companies/jimini) | 2026-09-19 |
| [Jimini homepage](https://jiminihealth.com/) | 2026-09-19 |
| [Jimini company page](https://jiminihealth.com/company) | 2026-09-19 |


## Author disclosure

I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.