Human Handover

Human Handover is a controlled escalation in which an ONOXIA bot conversation is taken over by a human support agent, with full message history and visitor context transferred to the agent's portal session.

Purpose

Bots are great at the median question. They are not great at angry customers, unusual edge cases, or decisions that need a person to sign off. Handover gives the bot a graceful exit and gives the customer a real human, without losing context.

Scope

Available on the Enterprise plan. Applies to chat conversations only (the email agent does not currently support handover). A handover preserves the conversation, the language, the visitor identifier (hashed), and the active Persona — the agent inherits whatever the bot was told.

Components

  • Trigger — visitor-initiated ("speak to a human") or bot-initiated (low RAG confidence, refusal-list topic, sentiment threshold).
  • Routing — round-robin across assigned agents, or by team queue.
  • Portal — agents work in onoxia.nz/support, see live conversation, send messages directly.
  • Timeout — if no agent claims within the configured window (default 5 min), the bot resumes.
  • Audit — every handover is logged with reason, agent, duration, and outcome.

Outputs

  • A live human conversation when the bot cannot help.
  • A clean transition for the visitor — no "please repeat your question" loop.
  • Capacity planning data: handover rate per persona, per site, per topic.

Relationships

Human Handover is triggered by a Persona Configuration when RAG confidence falls below threshold or a refusal-list match occurs.

Authority

Defined by OCENOX LTD.

Version

1.0 — 2026-05-22

Istilah terkait

  • Persona Configuration — A per-site bundle that defines who the bot is, how it speaks, which tools it may call, and what it must refuse. Personas sit between the visitor's question and the LLM, shaping the system prompt and tool registry for every conversation on that site.
  • Retrieval-Augmented Generation (RAG) — A pattern that retrieves passages from a private knowledge base before an LLM generates an answer, so responses cite your own documents instead of fabricating facts.