Data and AI

Data and Qanivo AI

Qanivo AI is presented to customers as one governed capability. Model output is a proposal; Qanivo-owned policy and action controls decide what may continue.

Response modes

  • Off disables customer-response generation for the applicable scope
  • Controlled generates a candidate that a permitted person reviews and deliberately sends
  • Autonomous is off by default and may send only eligible private replies after every governance and runtime gate passes

What remains human-authorized

  • Public comments and review replies require their accepted human-review boundary
  • Organic publishing requires approved content and schedule authority
  • Campaign activation and spend, audience, destination, tracking, and conversion changes require explicit human authority
  • Autonomous spending increases are prohibited

Data boundaries

  • Knowledge must be eligible, Published, Workspace-scoped, and traceable before it may support customer-facing AI output
  • External payloads, attachments, provider data, retrieved content, and model output remain untrusted until validated
  • Required evidence must be persisted before a sensitive customer-facing AI action continues
  • Provider fallback cannot weaken policy or broaden context

This website

The landing-page request form stores the submitted business contact fields in Qanivo’s first-party marketing lead store and may forward the same bounded record to a separately configured HTTPS lead endpoint. Marketing interaction events use an anonymous session identifier and exclude form values.