Beachfront Hospitality Group

The client is a beachfront hospitality brand with restaurant, beach and lounge or VIP experiences, and it receives a high volume of WhatsApp messages from guests asking about opening hours, reservations, availability, VIP tables, private events, policies and complaints. Without automation the team had to read every message, decide what the customer needed, check the right system and reply manually. Many guests ask similar questions about opening times, venue options, reservations, cancellation policies and general information, and those repetitive messages consumed time that belonged in higher-value conversations. Booking requests required understanding the guest count, date, preferred area and venue before availability could even be checked, which slowed responses during busy periods. VIP tables, bottle service, private events and large group enquiries are high-value opportunities, and missing or slowly handling those messages costs real bookings. Complaints are sensitive and cannot be treated like a normal FAQ message; they need a polite response and a clear path for team review. Guests also message in English or Spanish, so staff need to reply in the language the customer used.

The workflow runs inside n8n across nine steps, connecting WhatsApp, AI, Google Sheets, CoverManager and Gmail. It starts when a guest sends a message, and a validation step confirms the message includes a real sender before continuing. An AI classifier then identifies the language, a confidence level and the intent, with the main categories being FAQ, check availability, VIP enquiry, complaint and other general replies. A code step cleans and standardizes the AI result into a predictable structure containing language, intent, extracted details, the customer message and the chat ID, so downstream routing has something reliable to read. The message is routed to the correct branch: FAQ questions go to the FAQ agent, booking requests to the availability branch, VIP enquiries to the VIP branch, complaints to a handoff response, and small talk to a direct reply. FAQ questions are answered from the venue FAQ sheet, selecting the most relevant answer and replying in the same language as the customer. Booking requests use CoverManager tools for venue options, table availability and reservation actions, and can tell whether the guest is asking about the beach, restaurant or lounge and continue the conversation. VIP requests go through an additional validation step, and qualified enquiries trigger an internal email notification plus a professional acknowledgment to the guest. Complaints receive a polite holding response and are marked for escalation, keeping sensitive issues under human control while still replying immediately. Conversation memory keyed on the customer WhatsApp number lets the AI remember earlier details such as venue choice, date or party size.

A WhatsApp message from a guest is received and validated, with the language detected as English or Spanish and the intent classified as FAQ, check availability, VIP, complaint or other. A clean structured payload is created for downstream routing. FAQ questions are answered from the Google Sheets knowledge base, and booking and availability requests are routed to the CoverManager booking agent, which identifies venue context for beach, restaurant or lounge and VIP requests. VIP enquiries are validated and escalated to the internal team by email, with acknowledgment messages sent to the guests. Complaint messages are answered with a polite handoff response and flagged for team review rather than resolved automatically, which is the correct treatment for sensitive issues. General greetings and small talk are answered quickly without unnecessary routing, and final responses are delivered back through WhatsApp. The result is faster guest answers, fewer repeated manual replies, protected high-value VIP enquiries and complaints kept under human review.

Client retains ownership of the WhatsApp Business number, the Google Sheets FAQ knowledge base, the CoverManager account and venue data, the Gmail escalation address, the AI classifier prompts and confidence thresholds, the intent definitions, and every workflow and memory configuration.