Hotel AI Receptionist
WhatsApp guest assistant
- Status
- Shipped
- Role
- Automation engineer
- Source
- Private
Overview
An AI receptionist for a hotel, running on n8n and reachable on the channel guests already use. It answers questions about bookings, availability, and pricing by reading the hotel's live data rather than a static FAQ.
Conversations hold context across turns, and voice notes are transcribed automatically, so a guest can send a rambling audio message and get a useful answer back.
The problem
Front desk staff spend a large share of their shift answering the same handful of questions over WhatsApp — is my booking confirmed, what is the rate for those dates, what time is check-in.
Answering them well requires live data. A chatbot that quotes stale prices or cannot see a reservation creates more work than it removes, because every answer then has to be verified by a human anyway.
What was built
- Real-time MySQL integration for guest and booking lookups, so the assistant answers from the reservation system of record.
- Dynamic pricing synchronization via Google Sheets, which is where the hotel already maintains rates.
- Redis-backed load balancing that routes conversations across multiple Gemini models, keeping response times stable under bursts.
- Session memory so multi-turn conversations resolve pronouns and dates naturally instead of restarting each message.
- Automatic transcription for audio voice messages, which is how a significant share of guests actually write.
Tech stack
n8n
WhatsApp
MySQL
Redis
GeminiOutcomes
- Routine booking, rate, and availability questions are handled without front desk involvement.