your business
Upload your FAQs, help docs, policies, and SOPs, or scrape your site. Luran retrieves the most relevant passages to answer each question, and follows the rule you set when confidence is low, answer, clarify, refuse, or escalate.
Answers from your knowledge, not the open internet.
Upload your FAQs, help docs, policies, and SOPs, or scrape your site. Luran chunks and indexes everything, then retrieves the most relevant passages to answer each question.
- Choose your strictness. Strict, balanced, or flexible grounding modes.
- Low-confidence behavior. Answer, ask a clarifying question, refuse, or escalate.
- Stay on scope. Set allowed and blocked topics so the agent stays in its lane.
It sounds like you, and knows when to get a human.
Configure tone, greeting, formality, and emoji use. Set escalation rules, and decide when to collect a name or email.
Brand voice
Tone, greeting style, formality, emoji use, response length, and custom instructions.
Confidence-aware escalation
When unsure, it follows your rules, optionally confirming and collecting contact details first.
Pre-chat capture
Collect a name or email before chat or before escalation. Your call.
A clean handoff, every time.
Escalations become tickets with an AI-written summary and an inferred priority. Your agents take over the conversation and reply live; the customer sees a seamless transition.
- Ticket lifecycle. Open, in progress, resolved, closed. Priorities from low to urgent.
- Real-time over WebSocket. Typing indicators, agent presence, full message history.
- Unassigned queue. So nothing slips through the cracks.
Does it work without training data?
It needs your knowledge to answer well, that's the point. The more you add, the better it gets.
What happens after hours?
The AI answers 24/7; escalations queue as tickets and notify your team.