Compliance & patient safety
Trust, engineered in.
Not bolted on.
HIPAA compliance and a SOC 2 Type II audit in progress as the floor, not the ceiling. The clinician stays in control. Crisis indicators escalate to a licensed human. Every agent decision is auditable, end to end.
Ethics & privacy
Privacy by design.
Six commitments govern every model we train, every feature we ship, and every patient interaction the platform touches.
PHI minimization
We collect the minimum PHI needed for the clinical task, and never use patient data for model training without explicit, revocable consent.
Fairness on a cadence
Bias is evaluated on a published cadence across demographic groups. Drift is investigated before it reaches a clinician's screen.
Reasoning you can read
Every recommendation cites the source utterance, chart finding or policy clause behind it. The clinician sees the evidence, not a black box.
Clinician always in control
The platform drafts, suggests and proposes. A licensed clinician reviews and signs every care-impacting decision before it reaches a patient.
Patients retain the pen
Patients can review what was captured about them, request correction, and opt out of algorithmic decision-making at any time.
Training data provenance
Training data is sourced, documented and reviewed. An anonymous concern channel lets clinicians flag model behavior they don't trust.
Compliance posture
HIPAA, an active SOC 2 Type II audit,
and a BAA on day one.
Built to the frameworks healthcare compliance teams actually require, with controls monitored continuously in real time, not refreshed once a year.
Health Insurance Portability & Accountability Act
HIPAA
Administrative, physical and technical safeguards in place. BAAs are executed with every covered entity partner before a single record moves.
Service Organization Control
SOC 2 Type II (in progress)
Audit in progress against security, availability, processing integrity, confidentiality and privacy, evaluated over a continuous observation period, not a point in time.
Business Associate Agreement
BAA-grade by default
Every customer engagement begins with a BAA. Lawful use, disclosure and safeguarding obligations for PHI are settled before the first session.
Always-on control verification
Continuous monitoring
Continuous monitoring watches our controls in real time. The trust posture is verified continuously, not refreshed once a year.
Enterprise-grade
Privacy at
every layer.
Encryption, minimization and residency controls protect patient data from session to sign-off.
Encrypted end to end
PHI is encrypted in transit and at rest, with continuous integrity checks. The posture is BAA-grade, not bolted on.
Customer-owned encryption
Encryption keys belong to the customer. We never hold a copy that would let us read PHI outside the scoped clinical task.
PHI minimization
Identifiers are minimized before any record leaves the clinical context for analytics or quality improvement; sensitive fields are tokenized at the data layer.
Data residency
Customer data stays in the region the customer specifies. Cross-region movement requires explicit authorization, logged at sign-off.
Zero-trust by default
Every request is authenticated and authorized regardless of network. No implicit trust; identity is verified at every layer.
Auditable end to end
Every agent decision carries its evidence and its approver, reviewable by your compliance team at any time.
100%
of care-impacting outputs behind clinician sign-off
0
PHI used for model training without explicit consent
Every
agent decision auditable, end to end
Patient safety
Clinician always
in control.
The platform drafts and proposes. A licensed clinician reviews and signs. Crisis indicators escalate to a licensed human with a warm handoff, never to an automated reply.
Clinician sign-off, always
Every care-impacting output is reviewed by a licensed clinician before it reaches a patient. Model updates require clinical sign-off before they reach the visit.
Crisis escalation
Indicators of self-harm, suicidal ideation or acute distress trigger immediate escalation to a licensed clinician, with warm handoff protocols, never an automated reply.
Clinical safety guardrails
Hard guardrails sit between the agent and the chart. The platform never overrides a clinician, and when in doubt, it escalates to one.
Send this page to your
compliance team.
We share our SOC 2 Type II audit status, penetration test summaries, BAA templates and our control register with your compliance team on request, typically within one business day.
Safety runs through the product.
The same guardrails you read about here are the ones working in every session.
Watching quietly
All clear this session
Your plan, one tap away
Safety
Risk detection, escalation paths and compliance, the product pillar behind this page.
“What does the first minute feel like?”
Maria J.Warmth returning, noted
Co-therapy
Where the guardrails live day to day: Medera in session, clinician in control.