Catch what slips through the gaps between visits.
elli runs validated mental-health screening as light, conversational check-ins — then escalates safety signals straight to your care team. No app. No portal. No patient passwords.
The most important signals show up between appointments.
Primary care runs on an episodic snapshot model. Depression, anxiety, and alcohol risk build gradually in the long silence between visits — and screening, when it happens at all, is manual, inconsistent, and squeezed into a few crowded minutes.
Screening is inconsistent
Paper questionnaires at the visit, done if there’s time. Coverage is patchy and hard to repeat the same way twice.
The gap goes unwatched
Between visits, change accumulates unseen. The clinic is effectively blind until the next encounter.
Signals arrive late
By the next appointment, an early warning has often become a crisis. Care turns reactive instead of preventive.
A calm check-in that closes the loop.
From a secure link to an acknowledged safety alert — frictionless for the patient, structured for the clinic.
Secure link
A patient gets a single-use link by email or text — no account, no password. A quick date-of-birth check is the only gate. English or Spanish today.
A short conversation
elli asks a few questions in plain language — by text, email, or voice. Validated instruments — PHQ-2 → PHQ-9, GAD-7, AUDIT-C, plus condition-specific modules like C-SSRS and MDQ — delivered as a check-in, not a form.
Deterministic scoring
Responses are scored by exhaustively-tested clinical logic — the same way every time — and tracked for change across check-ins.
Safety escalation
A safety signal triggers an in-flow 988 response and a real-time alert to your on-call team by email and text, with timed re-escalation until it’s acknowledged.
Built to clinical-grade standards — not demo-grade.
Validated instruments
PHQ-9, GAD-7, AUDIT-C, plus condition-specific modules — C-SSRS graded suicide-risk and MDQ bipolar screening — delivered unedited to preserve clinical validity.
Deterministic clinical core
Scoring, branching, and safety triggers are plain, exhaustively-tested code. No AI in the safety loop.
Real-time safety escalation
In-flow 988 crisis response, acknowledged alerts, and timed re-escalation from on-call to fallback.
Multilingual, multi-channel
Check-ins run over text, email, or voice, in English or Spanish today — built to add languages and channels without reshaping the flow.
HIPAA-grade data handling
Append-only audit trail, row-level tenant isolation, and data minimization by default.
Multi-tenant for organizations
Every clinic isolated at the database level — built for multi-clinic groups from day one.
EHR-connected, not just EHR-shaped
US-Core-shaped Patient, Observation, and QuestionnaireResponse resources internally, plus a live Epic appointment sync and Epic chart-note write-back for completed screenings.
For the teams carrying primary care.
Independent practices
Primary care & internal medicine groups that want consistent screening without adding staff time.
Community health centers & FQHCs
High-volume settings where longitudinal tracking is hardest to sustain by hand.
Multi-clinic orgs & ACOs
Networks that need isolation, audit, and consistency across every site.
Designed for the team that has to vet you.
elli is pre-launch: live software on real infrastructure, running on synthetic test data while BAAs are finalized. No real patient data — today or in this site.
Three things finally line up.
Policy
Medicare’s 2026–2028 schedules reward continuous early detection through APCM — funding the work from day one.
Capability
Conversational AI can finally sustain warm, frequent, plain-language check-ins at scale.
Need
Primary care is too overloaded to screen consistently by hand. elli sits exactly in that gap.
See elli with your own workflow.
Book a 30-minute demo. We'll walk the full loop — enrollment, secure check-in, scoring, and the safety escalation — on synthetic data.