Case study: how a Langkawi beach resort handled 2x booking volume with an AI WhatsApp concierge

The business

A 32-room boutique beach resort on the northern coast of Langkawi, Kedah. Family-run, opened in 2014, primarily a direct-booking and Agoda/Booking.com business with a small wedding and events side. Two front-of-house staff during the day, one at night, and the owner handling the WhatsApp enquiries that came in around the clock.

This is the case study for the AI chatbot side of what we do — a small, opinionated build for a Kedah hospitality business that did not need a "chatbot to look modern." It needed to stop losing bookings because the owner was asleep.

The problem they were actually trying to solve

A peak-season booking window that the resort was sleepwalking through. Langkawi's high season (December to March) was when enquiries tripled, and the resort's WhatsApp inbox became a black hole. The owner — who also handled the night shift on the front desk twice a week — would wake up to 30+ messages, half of which were booking requests that had already moved on to the next resort.

Three leaks, each costing real money every high season:

• Response time on WhatsApp: 4 to 8 hours. During peak season, often the next morning. Booking intent drops sharply after 2 hours of no reply.
• Quote consistency: shaky. The two front-of-house staff had slightly different ways of quoting (one included breakfast, one didn't, until corrected). The owner was the only one who knew the actual room rates and seasonal offers.
• After-hours enquiries: 70% of total. Langkawi attracts a lot of European and Singaporean bookers, who message in their timezone. By the time the resort replied, the booking was gone.

The owner had tried two off-the-shelf WhatsApp business tools in 2024. Both worked technically but neither "got" the resort — they couldn't quote a specific room type, didn't know the seasonal offers, and pushed leads to a generic booking link that the owner had to maintain separately. The owner kept both tools and disabled them.

What the owner said on the first call: "I need something that actually knows my resort, quotes correctly, and books the room while I sleep." That sentence is the brief.

What we built — and why this approach

A small AI WhatsApp concierge with three parts, sitting on the resort's existing WhatsApp Business number:

1. A trained AI concierge that knows the resort's 5 room types, the seasonal rate card, the breakfast and transfer add-ons, the wedding packages, and the cancellation policy. It quotes, books, and confirms — all inside WhatsApp. The booking is pushed to the resort's existing PMS (Cloudbeds in this case) via a small API hook.
2. A human-handoff rule that escalates anything complex — wedding enquiries over 20 pax, corporate retreats, special requests like a private dinner — to the owner in under 30 seconds, with the full conversation history.
3. A morning summary the owner reads at 7am: yesterday's enquiries, bookings made, handoffs escalated, and the AI's confidence score on each conversation.

Why not the off-the-shelf WhatsApp tools? Because they tried to be generic — any business, any industry. The concierge we built is narrow: it knows this resort's 5 rooms, this resort's rates, this resort's policies. If the resort opens a second property, we train a second concierge. That's the architect, not coder difference — we spec'd the model to do exactly what this resort needs, not what a SaaS vendor wanted to sell.

How the operation changed

Three concrete changes inside the first 8 weeks:

• The owner stopped waking up to 30 unanswered messages. The AI handles 80% of the enquiry volume end-to-end. The remaining 20% — complex bookings, complaints, special requests — get handed to the staff with full context. The owner sleeps through the night, more or less.
• Quote consistency went from "shaky" to "perfect." Every quote includes breakfast, every quote names the room type, every quote references the cancellation policy. The two front-of-house staff stopped arguing about which rate card was current.
• After-hours bookings went from 0% to 38% of total. The AI's biggest contribution. A Singaporean couple messaging at 11pm SGT gets a quote, a booking link, and a confirmation in under 90 seconds. The resort's previous 0% after-hours capture is now its fastest-growing channel.

The build is now in the second phase — multilingual (the AI now handles English, Bahasa Malaysia, and Mandarin, the three languages the guest mix actually uses) and a small upsell layer for the in-house spa and airport transfer services. That's the business automation layer that pays back on a different line of the P&L.

The numbers, eight weeks in

We don't do client testimonials. We do numbers. The resort's view, eight weeks after the AI concierge went live (covering the December-January peak):

• WhatsApp response time: 4-8 hours → under 90 seconds, 24/7. The single biggest behaviour change. The resort's average time-to-first-reply is now 38 seconds.
• Bookings made via AI: 38% of total bookings during the period. 73 actual bookings across 8 weeks, average length of stay 3.4 nights, average daily rate RM 385. Roughly RM 95,000 in revenue that the resort's previous "we'll reply in the morning" approach would have lost.
• Conversion rate (enquiry → booking): 18% previously → 34% with the AI. Doubled, mostly driven by response time and quote consistency. The numbers line up with industry research on WhatsApp booking conversion.
• After-hours bookings (10pm-8am): 0% previously → 38% of AI bookings. The biggest single revenue contribution. The Singapore/European guest market the resort couldn't reach before is now reachable.
• Front-of-house hours on WhatsApp: ~5 hours/day → ~1.5 hours/day. The recovered time goes into actual guest experience — welcome drinks, tour recommendations, the small touches that earn 5-star reviews.

Total first-phase cost: RM 22,000 to RM 45,000, spread across 8 weeks. The WhatsApp Business API + model hosting is the main ongoing cost; the rest is model training, the PMS integration, and the 90-day post-handover review.

Payback: 4 weeks. The first month of peak-season AI bookings alone covered the entire program cost.

What it would cost for your business

We don't publish a price sheet, because the price depends on what you're running. Three honest bands for a Kedah or Perak hospitality business:

• RM 18,000 – RM 35,000 (one-off) + RM 500 – RM 1,500 / month. A first-phase AI WhatsApp concierge for a 10 to 40-room Langkawi, Cameron Highlands, Penang or Ipoh hotel with one WhatsApp number, 1 to 3 room types, and an existing PMS. Includes model training, API integration, and 90 days of tuning.
• RM 35,000 – RM 70,000 (one-off) + RM 1,000 – RM 2,500 / month. A multi-channel concierge for a 40 to 120-room hotel covering WhatsApp + website chat + Instagram DM, multilingual (English + Bahasa Malaysia + Mandarin), with PMS, payment, and a small upsell layer.
• RM 70,000 – RM 140,000 (one-off) + RM 2,500 – RM 5,000 / month. A multi-property concierge for a small hotel group with 2 to 4 properties, 1 concierge per property, central reporting, and a handoff layer to the reservations team.

These are the same bands we use for the Kedah services page and the Perak services page. Every engagement is priced for the build, not for the hours we put in.

What this looks like for your business

If you run a Langkawi resort, a Penang boutique hotel, a Cameron Highlands stay, or any Malaysian hospitality business that takes bookings on WhatsApp and is losing them because the reply is too slow — this kind of build is the right first move. We don't sell a generic WhatsApp chatbot. We build a concierge that knows your rooms, your rates, and your policies, and books them 24/7.

The next step is a one-hour call, no slide deck. You tell us your property, your room types, your current booking flow. We tell you whether an AI concierge is the right answer, or whether a cheaper off-the-shelf tool will do. We won't pitch you either way. Get in touch and we'll set up the call.

If your bottleneck is the F&B side rather than the front desk, our Ipoh kopitiam case study shows the AI front-desk build for a 3-outlet F&B group. And if your bottleneck is in the back office — stock, orders, AR — our Kedah auto-parts case study shows what the custom software side of the same engagement looks like.