Responsible AI
How QotBot is designed to behave — and how teams maintain control over what it does.
Responds only from approved knowledge
QotBot answers questions using the knowledge base and playbooks your team provides. It does not generate responses from general training data or external sources.
Explicit escalation — no dead ends
Every conversation has a configured path to a human. QotBot escalates when it cannot answer, when a topic is flagged as sensitive, or when a customer explicitly asks.
Does not replace clinical judgment
In healthcare settings, QotBot handles operational communication only. It does not diagnose, prescribe, or offer clinical advice. Sensitive health questions route to qualified staff.
Consent before outreach
QotBot sends outbound messages only to contacts with recorded, explicit consent. STOP, CANCEL, and UNSUBSCRIBE keywords are honored immediately and logged.
Prompt and response version logging
Every AI response is linked to the prompt version that produced it. Teams can audit why QotBot said what it said at any point in time.
Configurable guardrails
Teams define which topics QotBot should never attempt to answer. Out-of-scope topics trigger a fallback — not a guess.
Human review of outcomes
QotBot provides dashboards and escalation logs so teams can identify gaps, retrain knowledge, and adjust escalation rules based on real conversations.