The client
A small luxury resort group with three boutique properties on the Andaman coast. Around two hundred staff in total. Front-desk teams were stretched on guest enquiries, especially during peak hours and overnight shifts.
The problem
A high-end guest expects a quick, polite, accurate answer to questions like “what time does the spa open today” or “can someone arrange a longtail to the next island tomorrow morning”. Front desks handled these messages over Line and WhatsApp on top of everything else. Quality varied. Late-night messages sometimes went unanswered until morning, which is not the standard a five-star property wants to hold.
The approach
The build started with the property’s own knowledge base, structured into Airtable: every restaurant’s hours and dress code, every activity’s pricing and availability windows, every transport option, the full list of room amenities. n8n watched the Line and WhatsApp inboxes. An LLM read incoming messages, looked up the answer in the knowledge base, and replied in the resort’s voice, in the guest’s language.
The system stayed strictly inside its knowledge base. Anything ambiguous (a complex booking, a complaint, an unusual question) got handed straight to the human concierge with full context. Guests never noticed the handover except as a slight tone shift.
The result
About seventy percent of incoming guest messages got handled inside thirty seconds, around the clock, without the front desk getting involved. Guest satisfaction on responsiveness scored its highest year on record. The human concierges were freed for the requests where they actually add value: arranging a private boat at sunset, finding a last-minute restaurant booking on a public holiday.