Trust & safety

Built for healthcare from the compliance layer up

Rivvi is HIPAA-eligible and SOC 2 Type II. A BAA is signed during signup, before any patient data moves. TCPA opt-out is honored on every outreach path, and clinical guardrails run on every conversation.

  • HIPAA-eligible
  • SOC 2 Type II
  • BAA at signup
Clinical guardrails

Rivvi does not diagnose, prescribe, or override clinical judgment

The design preserves clear boundaries. Rivvi's AI does not diagnose, prescribe, or override clinical judgment. Clinical judgment stays with clinicians and choice stays with patients; the platform creates capacity to bring more people into an informed discussion with the right care team at the right time.

Each organization configures its own policies, escalation rules, access controls, and audit trails for its workflows. Every action Rivvi takes is personalized, governed, and auditable, and each transition can be authorized, logged, and escalated. Rivvi's platform is HIPAA compliant and SOC 2 Type II, with business associate agreements available, and supports deployment without requiring a new electronic medical record integration.

Compliance handled before anything runs
  • HIPAA
  • SOC 2 Type II
  • GDPR
  • BAA included
  • TCPA compliant

Security and compliance

Certifications

SOC 2 Type II, HIPAA BAA, TCPA in production

Rivvi maintains SOC 2 Type II. Request the current report at security@rivvi.ai. A HIPAA Business Associate Agreement is executed during signup, before any record moves. PHI stays inside HIPAA-eligible infrastructure and never trains third-party models.

TCPA consent capture and opt-out run on every outreach path and have been in production for years. GDPR applies where our 8-country footprint processes personal data of individuals in the EEA — data residency and subprocessors are listed below and in the Trust Center.

Data & privacy

PHI stays inside HIPAA-eligible infrastructure

Every model call runs on HIPAA-eligible infrastructure covered under a BAA. Protected health information never leaves that boundary and never trains third-party models — a hard rule enforced at the platform layer, not a setting you have to find.

Access is org-scoped end to end: your data is isolated to your organization, and every read and write is scoped to it. SOC 2 Type II controls govern how that access is managed and audited. Primary processing runs on AWS (US regions) with telephony and messaging subprocessors under BAAs; see trust.rivvi.ai for the current subprocessor list.

Governance

Your policies, enforced on every conversation

Define your rules once in plain language; they apply to every conversation immediately, consistently, with zero drift.

  • ESCALATION

    Escalation rules

    Route any conversation that needs a person to the right member of your team, in seconds, with full context.

  • CONSENT

    Consent & opt-out

    TCPA opt-out is honored on every path; consent state is tracked and respected before any outreach runs.

  • TIME

    Time-based rules

    Quiet hours and calling windows are enforced automatically, per your policy and the patient's locale.

  • ACCESS

    Role-based access

    Who can see and do what is governed by role, org-scoped, and auditable.

Safety

Guardrails that protect patients

AI in healthcare can't afford to get it wrong. Every conversation runs inside clinical guardrails with real comprehension, not keyword matching.

  • CRISIS

    Crisis detection & escalation

    The agent recognizes distress and crisis signals in context and escalates to a trained responder immediately, not at the end of a script.

  • SCOPE

    Medical-advice guardrails

    The agent stays inside its scope: it never gives medical advice, and it hands off cleanly when a question needs a clinician.

  • VULNERABLE

    Vulnerable-population protections

    Extra care for at-risk patients is built into how the agent listens, responds, and decides when a human should take over.

  • RECORD

    Automatic documentation

    What happened, what was said, and what was escalated is documented automatically, so nothing depends on memory.

Evidence

See the controls, not just the claims

Our Trust Center carries the current certifications, subprocessor list, and security documentation your compliance team will ask for. Security questionnaires and BAA requests are handled there.

The BAA is e-signed during signup, so you're on real patient data on day one — inside the same posture described here, not a scripted demo sandbox.

See it run on your own patients

Spin up a workspace, sign your BAA, and launch a real program today. No sales call, no integration project.