ethics
AI without ethics is liability. We engineer guardrails before we engineer features — and publish how each one works.
01
Human in the Loop
No client-facing output is delivered without an attorney-supervised review pass.
02
Zero Raw Outputs
Models never speak directly to clients. Every response is filtered, cited, and audited.
03
Edge Sovereignty
Inference runs on dedicated edge nodes within your jurisdiction. No third-party logging.
04
Disclosed Automation
Callers are told they are speaking with an AI receptionist. Consent is logged.
05
Conflict-First Intake
Conflict checks run before any matter intake is recorded — never after.
06
Right to Human
Every interaction has a one-press path to a real human. No dark patterns.
Read the full charter
Our public ethics charter governs every model deployment, every prompt, and every escalation path.
Talk to our Ethics Lead