Collection
Aura AI
The AI agent: enabling it, teaching it, deciding where it may answer, and measuring what it does.
Articles
9 in total- Enabling Aura AIThe one switch that creates your agent, what it turns on, how to test it, and what disabling does and does not delete.
- Giving the agent a personalityThe seven settings on the Agent page — tone, persona, brand voice, role, goals, company info and glossary — and which settings live elsewhere.
- Training data: what the agent knowsThe indexed-document list, the three ways to add sources, and what re-indexing and deleting a document actually do.
- Agent Assist in the inboxThe Aura AI tab beside a ticket: drafting a reply, asking a private question, reading the citations, and the four AI views in the sidebar.
- Autonomy: where the AI may answerOff, Suggest and Auto, how a rule's scope decides which one wins, and the keep-a-copy-on-email default.
- Auto-resolveLetting the AI reply and solve a ticket unaided: the confidence threshold, the allowlist that must be filled in, the daily cap and the guardrails you cannot switch off.
- Ticket categorisationLetting the AI assign a ticket type to every inbound email, why it declines rather than guesses, and how to read the decision log.
- Measuring Aura AI's performanceThe date ranges, the ten headline measures in three bands, and the four lists that tell you what to fix next.
- Workflows: building the agent's flowThe node library, the twelve node types, how edges and routes work, and why saving replaces the whole graph.