Starter pack · Hotels & travel
A hotel that answers guests in any language on any channel, and knows its numbers each morning
Four AI agents for small and mid-sized hotels, resorts, hostels and villas. They answer booking and enquiry chat in Thai, English and Chinese, look after reviews on the OTAs, take care of guests from before arrival through to check-out, and give the manager occupancy, rate and revenue every morning. Pre-assembled, live in 2–3 weeks.
Work in
Hotels & travel+ human approval
To the customer
A person approves before it ever reaches the customer· drag to rotate
A fit if you…
- Hotels or resorts of 20–150 rooms across 1–3 properties, with a handful of front desk staff answering LINE, WhatsApp, Facebook and OTA chat by hand, day and night
- More foreign guests than the team can comfortably answer in English or Chinese, OTA reviews left unanswered, and a score drifting down
- A manager who learns occupancy and revenue at month end, too late to move a rate or run a promotion on the empty nights, and who does not want to pay for an Audit first
A day with this pack running
- 07:00The agent sends the manager the morning summary: tonight's occupancy, rooms left, average rate per room, yesterday's revenue, how many guests check in today, and new reviews still unanswered
- 09:30A Chinese guest asks about rooms for the long weekend on WhatsApp. The agent answers in Chinese, offers what is free at the hotel's own rates, and sends a direct booking link. A request for a special discount goes to the front desk to decide
- 11:00Pre-arrival messages go to tomorrow's guests: details to fill in, rough arrival time, and an offer of a transfer or a room upgrade on the terms the hotel set
- 15:00A guest in-house messages on LINE for extra towels and asks about restaurants nearby. The agent sends the job to housekeeping and recommends from the hotel's own list. A complaint about the room reaches the duty manager at once
- 18:00Three new reviews arrive from the OTAs. The agent drafts replies in the hotel's tone, in the guest's language. Good scores it can answer itself; negative reviews go to the manager to read and approve before posting
- 21:00Thank-you messages and a feedback request go to today's check-outs. Happy guests are invited to leave a review; unhappy ones are passed to the manager to call back before it becomes a public review
The output you get
Guest
Can I check in at 11am and keep luggage?AI
Luggage storage is open from 09:00. An 11:00 early check-in depends on availability — we will confirm again on the morning you arrive.Answers in Thai, English and Chinese · refunds and complaints go to the manager
An illustrative example — the names and figures are invented so you can see the shape of the work; the real thing is built around your business.
Example: Hotels & travel
LIVE DEMO1 · Work arrives
2 · The AI agent does it
- Booking chat in every language
- OTA reviews and replies
- Guest care before, during, after
- Daily occupancy, rate, revenue
3 · You just approve
Discounts outside the terms, group bookings over 5 rooms, and special requests that affect rooms or staff go to the front desk to decide before anything is confirmed.
Result
-70%
chats staff have to answer themselves
What is in the pack
4 AI Agent already assembled
Answers questions on availability, rates, facilities, getting there and booking terms, in Thai, English and Chinese, 24 hours a day on LINE, WhatsApp, Facebook and OTA chat. It checks availability in the real booking system and can walk a guest through to a direct booking in the same conversation.
Discounts outside the terms, group bookings over 5 rooms, and special requests that affect rooms or staff go to the front desk to decide before anything is confirmed.
Collects new reviews from every OTA and map listing in one place, drafts replies in the guest's language in the tone the hotel set, and summarises weekly what guests praise and complain about most, so the team can fix the cause.
Low scores, reviews that name an individual member of staff, and anything about safety have to be read and approved by the manager before the reply is posted.
Sends pre-arrival check-in and offers upsells, such as a room upgrade, breakfast or a transfer, on the terms the hotel set. During the stay it takes requests and passes them to the department concerned. After check-out it asks for feedback and invites a review only from guests who were happy.
Complaints during the stay, compensation and room moves reach the duty manager at once. The AI does not decide compensation or refunds.
Pulls the numbers from the booking system and the OTA channels every morning and summarises occupancy, average rate per room, revenue per available room, and the split between direct and OTA bookings, flagging unusually empty dates two weeks ahead so the manager sees them first.
The AI can suggest something like a weekday promotion, but moving rates and opening or closing a sales channel is the manager's decision alone.
All included
What is in the box
- LINE OA, WhatsApp, Facebook Messenger and the OTA inboxes the hotel uses, connected into one inbox with handover to staff
- A knowledge base of rooms, rates, facilities and what to do nearby, in 3 languages, that the front desk edits itself
- Pre-arrival, upsell, in-stay and post-check-out feedback messages, with the rules for what is offered to whom and when
- An approval screen for the manager: replies to negative reviews, discount requests, complaints and group bookings
- A PDPA data map, consent forms for guest data and check-in documents, a retention and deletion policy, and access logging
- A daily and weekly dashboard (occupancy, average rate, revenue, reviews, questions the AI could not answer) and 2 training sessions for the front desk and the manager
Go live
Live in 2–3 weeks
Step 1 / 3 · Collect the hotel's rules and set up PDPA
Week 1
- 1Week 1
Collect the hotel's rules and set up PDPA
Two to three hours with the owner, the manager and the head of reception to collect room types, rates and terms, the tone for review replies, the upsell list, and the line the AI must not decide on its own, plus the data map and the consent forms.
- 2Week 2
Connect the booking system and test on real chats
- 3Week 3
Go live channel by channel and tune
-70%
chats staff have to answer themselves
from the properties we have done, within the first month, because repeat questions on rooms, rates and directions are answered by the AI in every language
+0.3 to +0.5
average review score on the OTAs
usually visible in month 2–3, from answering every review and fixing guest problems before check-out
+12%
extra revenue per stay
from upselling room upgrades, food and transfers at pre-arrival check-in, measured against before go-live
Price and scope
One price, paid once, to install the whole pack, plus a small monthly fee for models and keeping the system running. All in, about the same as one part-time receptionist per month. Fixed scope as set out in the pack, with no hidden costs.
Not included
- Building a new booking system or PMS, building a direct booking website, and migrating everything out of the old system
- Automatic rate changes against competitors (dynamic pricing) and a Voice Agent for phone calls — both can be added in month 2–3
- Marketing, content, running ads, and managing the hotel's OTA accounts on your behalf
Where you go when you grow
Once you run more than 3 properties, or want to connect to a PMS and a complex rate system, move up to Agentic Workflow Design and Implement & Coach built around your hotel, or add the Sell more and Customer service on every channel bundles, without starting over.
Safety in this business
- The AI does not confirm discounts, does not award compensation, does not refund, and does not post replies to negative reviews on its own. Anything that touches the hotel's money or its reputation is approved by a person first.
- Guest data such as passports, phone numbers and stay history falls under PDPA. We take consent before collecting, set how long data is kept and how it is deleted, and log who accessed what and when.
- Conversations and guest data are not sent out to train models. The model sees only the data that task needs, and you can choose a model that runs in Thailand or a Private LLM when the hotel wants it.
FAQ for Hotels & travel
Is the AI's Chinese or English really good enough? Will guests know it is AI?
It answers at a usable standard in all 3 languages, and we have a native speaker or a fluent member of staff check the core answers before go-live. Guests can see clearly that it is an automatic assistant, and can ask for a member of staff at any time — the handover comes with the conversation summarised in a language your team reads.
Will the AI answer a bad review itself and make things worse?
No. Low scores, reviews that mention staff, and anything about safety are drafted and then wait for the manager to approve before posting. The AI can post on its own only for good scores, in the tone the hotel set, and the manager can switch that off at any time.
We already run a booking system or PMS. Do we have to change it?
No. We connect to booking systems and channel managers that offer an API or can export data. If the old system cannot connect at all, the AI works from the availability sheet your staff update, and passes any availability question it is unsure about to a member of staff instead.
How safe is guest data, such as passports?
This pack comes with a data map, consent forms, a retention and deletion policy, and access logging. Guest data is not used to train outside models, the model sees only what the task needs, and you can use a model that runs in Thailand or a Private LLM if the hotel wants it.
How do we start, and what do we need to prepare?
Start with a free 30-minute call, no obligation, to see whether this pack fits your property. Then have the room and rate list, the channels guests message on, and 1–2 months of past chats and reviews ready. We handle the rest within 2–3 weeks.
Cases from similar businesses
Packs for other businesses
A free 30-minute call to see if the hotel pack fits you
Tell us how many rooms you have, which channel guests use most, and why reviews go unanswered. We will say straight out whether this pack is enough or whether it should be designed around you. No obligation.
Want to see it working first? Open the sample workspace for a business like yours — this exact team of agents is running in it. Sample data, no sign-up.