Pre-launch.

What this page will track when we go live

• Web (trellova.com)
  Marketing site uptime.
• Clinician app (app.trellova.com)
  BFF availability + median p95 latency.
• Patient app (patient.trellova.com)
  Magic-link + intake flow.
• Ambient capture (LiveKit + scribe)
  End-to-end session reachability.
• LLM pipeline (scribe → coder → PA)
  Per-step success rate + queue depth.
• EHR write-back
  Per-vendor success rate (TherapyNotes, SimplePractice).
• SMS / email delivery
  Twilio + Postmark delivery rates.
• Database + encryption
  RDS + KMS health (per-tenant DEK decrypt latency).

How we'll handle incidents

Our incident posture is documented before production exists, not after. For every category of failure that matters in a clinical workflow we have a runbook with explicit response targets, evidence-collection steps, and customer-notification templates — so when something does break, the path back is rehearsed.

• PHI breach (P0) — 5-min acknowledge target; counsel-led 60-day notice path per HIPAA 45 CFR §164.404.
• Scribe quality regression (P0/P1) — “stop the line” rule for any clinical-safety regression; eval-gate verification before deploys resume.
• Multi-region failover (P0) — RPO ≤ 5 min, RTO ≤ 60 min. Quarterly drill, automated CLI, drill report archived.
• Denial-rate spike (P1) — payer-side vs. tenant-side vs. Trellova-side classification, per-bucket remediation, customer re-draft path if we caused it.

Contact during an incident

Right now — and during production once it's live — hello@trellova.com reaches us directly. Production-prod will also post incidents here in real time with a per-region uptime panel.