Sage by Ellipsis Health – A Voice‑Enabled, Emotion‑Sensitive AI Care Manager

Ellipsis Health recently raised $45M in Series A‑1, backed by CVS Health Ventures, Salesforce, and Khosla Ventures, to expand Sage, its AI care management platform. Designed to fill critical gaps after patient discharge, Sage autonomously calls patients to:

• Clarify discharge instructions

• Track medication adherence

• Coordinate transport and care logistics

• Detect emotional distress with its “empathy engine” trained on two million clinical calls, adapting tone and escalating urgent cases ﹘ bridging the empathy gap in automated calls.

EA angle: Explore the architectural essentials—integrating Sage with EMRs, remote monitoring devices, secure workflows, and human‑in‑the‑loop escalation layers.

EA Integration Blueprint: Plugging Sage into the Enterprise

Use this Mermaid diagram to illustrate the key architectural domains:

Why it matters to EAs:

• Define APIs and HL7/FHIR pathways for seamless data exchange.

• Ensure privacy‑by‑design and auditability across voice logs.

• Architect “human‑in‑the‑loop” triggers to comply with clinical safety and regulatory standards.

Spotlight: Voice AI Across the Care Continuum

• Speedoc (Singapore) uses voice AI to triage home‑care patients, optimize visits, and predict clinical risks, but emphasizes AI as a productivity, not replacement, tool.

• In the US, voice AI automates insurer interactions and provides companionship for isolated seniors, boosting processing speed and social connection while keeping human oversight in place.

EA Insight: Voice agents must integrate with back‑end systems (insurance, scheduling, alerts) within governed architectures that balance automation, trust, and quality oversight.

Rapid‑Win Opportunities for Fractional EAs & Transformation Programs

1. Discharge Care Pilot

   • Target: Diabetes or post‑surgical patients

   • Deliverables: Sage integration with EMR & device data, KPI dashboard (readmission, adherence)

2. Home‑Care Triage Workflow

   • Architect layered voice/telemetry logic

   • Include human escalation and continuous performance monitoring

3. Integrated Voice‑Bot for Claims and Intake

   • A hybrid AI assistant for insurers/providers

   • Addresses automation of benefits verification, call documentation

EA Governance Snapshot: Key Domains & Metrics

Domain	Key Responsibilities	KPIs
Data & Integration	EMR, voice data, device telemetry integration via FHIR/HL7/APIs	% of successful syncs
Security & Privacy	Encryption at rest/transit, consent management, audit trails	Compliance: HIPAA/GDPR, % of escalated calls with logs
Human‑in‑Loop Strategy	Escalation logic, role clarity, triage thresholds	Escalation rate, resolution time
Error Handling & Bias	QA/testing of logic paths, sentiment accuracy	False-positive/negative rates, bias audits
Monitoring & Reporting	Real‑time dashboards with alerts + compliance logs	Care engagement, patient satisfaction

Key Takeaways for Fractional EAs

• High-impact, short engagements: Perfect fit for fractional EA—quick definition of integration scope, pilot execution, and governance design.

• Transformational outcomes: Reduces discharge risk, enhances patient engagement, and relieves care teams.

• Scalable architecture: Built modular and governed correctly, your EA design can be replicated across regions, specialties, or insurance lines.