Hotel AI training: a step-by-step guide

When booking season opens, the reception phone and inbox never stop: "Do you have a sea-view room for August 8-11?", "How much for two adults and one child?", "Is breakfast included?", "Is a late check-out possible?", "Is parking paid?". The same questions arrive separately via Booking, WhatsApp, Instagram DM, and email. Reception has to keep up with all of this while welcoming guests in person. Most of the time, they can’t; the "we’re arriving tomorrow, do you have a room?" message that lands at 11 PM waits until morning, and the guest books another property.

inTusell takes over this traffic. But to do so, it first needs to learn about your property, your room types, your prices and policies, and your way of talking. This article explains how a hotel or accommodation business trains the inTusell AI from scratch, step by step. This is the hotel/accommodation chapter of our sector-based "how to train your AI" series, and it is the cornerstone article of the series.

Quick answer

Why does training matter?

inTusell is not an off-the-shelf chatbot; it is an AI sales assistant and CRM that answers messages on WhatsApp, Instagram and web chat, books appointments and opens orders, and follows each customer through the sales pipeline. It is trained to behave like an experienced reservations advisor. A good receptionist doesn’t start on day one without knowing the room types, pricing policy, and house rules. Training ensures the AI does two things correctly: give the right information (room types, real prices, cancellation policy) and give it the right way (your property’s tone, your way of welcoming guests).

An untrained assistant either gives overly generic answers or guesses on a topic it doesn’t know. In accommodation, guessing is costly: a wrong nightly rate means disappointment at booking, made-up availability means double bookings and cancellations, and a wrong cancellation term means complaints and bad reviews. A well-trained assistant stops when it doesn’t know, says so honestly, and connects to the live team. This is the strongest trust message you can build. The purpose of training is not to make the AI talk a lot; it is to make it talk correctly and stay silent in the right place.

Who is it for?

This guide is for accommodation businesses whose incoming message traffic is dominated by reservation and room questions and who answer the same things over and over:

• City hotels, boutique hotels, holiday resorts, aparthotels, guesthouses, and villa-rental properties
• Teams receiving dozens of availability and price questions a day via Instagram and WhatsApp
• Properties that want to grow their direct (commission-free) bookings beyond channels like Booking and Airbnb
• Hotels in tourist areas that communicate in multiple languages with foreign guests
• Everyone overwhelmed by message volume, from a one-person reception to a separate reservations team

You don’t need a technical team. All of the steps below are completed from sections in the panel, without writing code.

The structure of training: 5 core steps + 2 continuous steps

Splitting the training into two groups makes your job easier. The first 5 core steps get the assistant up and running; these steps are the one-time part of the setup. The 2 continuous steps that follow (reviewing and turning fixes into rules, and opening channels) are the ongoing part that you sharpen as you use the assistant. You’ll find all of them in order below.

1. Upload the knowledge base

The foundation of training is the knowledge base (RAG). In the panel’s My business › Knowledge › Knowledge Base section, you upload your property’s information. Each time the AI generates a response, it searches these sources and uses only the information written there. This is the technical answer to the "don’t let the AI make things up" worry: the AI talks about whatever you hand it.

Supported source types:

Every file you upload is automatically chunked and made searchable with pgvector. When you upload your room types and a guest asks "Do you have connecting family rooms?", the AI finds the right section and responds.

What an accommodation business must upload: room types and capacities, room amenities (sea view, balcony, kitchen, air conditioning), breakfast/half-board/full-board terms, check-in/check-out times, cancellation and deposit policy, property details such as parking/pool/pets/children, surroundings and transport (airport transfer, distance to the beach), and the most frequently asked questions. Once these are in, the answers to the most frequent questions are ready.

An important note: when room equipment, amenities, or policies change, keeping the knowledge base current is up to you. When you upload the updated file, the AI uses the new information once the file is processed. The AI won’t present a policy from an old file as "current"; it knows whatever you give it, and it says what it doesn’t know.

2. Keep the room and price source accurate

In the first step you uploaded general texts (room types, amenities, and policies) to the knowledge base; in this step you focus on keeping your actual product backbone, the room and price information, accurate. There is an important accommodation-specific distinction here: there are two kinds of information, and they come from different places.

• General, unchanging information (room types, capacity, amenities, cancellation policy): goes into the knowledge base. Questions like "How many people fit in the suite?" or "Is breakfast included?" are answered from here.
• Date-specific availability and nightly rate: comes from the room reservation engine in the Stays module (beta; your sector must be Hotel & Accommodation and you need to enable the module). A question like "Is there a free deluxe room for July 12-15, and how much?" is answered by this engine. The AI takes the dates and party size, the engine calculates the price for those dates from your date-range rate plans, and allocates the room in one locked transaction.

This distinction is the heart of the "will the AI make things up?" question. The AI does not state a nightly rate or availability off the top of its head; it pulls these only from the reservation engine’s rate and availability query. For a date or room you have not fed the engine, it does not guess; it says it does not know and offers to connect to the team. Because allocation happens in one locked transaction, two requests coming through inTusell cannot take the same last room. There is no channel manager connection: rooms you sell on OTAs or by phone you record yourself in the availability calendar by lowering the quota or closing sales. When the module is off, the AI never sees the room tools and makes no availability promise.

The practical setup looks like this:

• Upload room types, capacities, and amenities to the knowledge base as a clean file; remove old/conflicting files.
• Enable the Stays module and define seasonal prices as date-range rate plans so the AI pulls the real price from there.
• Write conditions such as cancellation, deposit, and late check-out into the knowledge base so they are conveyed consistently in every response.

We covered how the room reservation engine is set up and how it works in a separate article: hotel room reservation automation. You can also see all the capabilities of the knowledge base on the solutions page.

3. Set the persona and tone

You taught the AI what to say in the first two steps; now you’ll define how it says it. There are two settings in the panel:

• Assistant name (ai_persona_name): The name it will introduce itself with to the guest. Most properties use a name aligned with the brand or a real reception/reservations advisor’s name.
• Tone (ai_tone): Whether it speaks warmly and invitingly, corporate and crisp, or sophisticated and calm.

The tone of a family holiday resort and a boutique hotel in the city center won’t be the same. The former might be warm, inviting, and full of "welcome"; the latter measured, elegant, and selective. A seaside guesthouse uses relaxed, everyday language. This setting is reflected in the AI’s message style on every channel; the tone in an Instagram DM stays consistent with the tone in web chat.

A multilingual note: The AI responds in whatever language the guest writes in. If you are in a tourist area, when a foreign guest writes in English, German, or Russian, you do nothing extra; the assistant detects the language and responds in it while preserving your tone. In accommodation, this is a practical way to avoid losing a direct booking.

4. Define the response rules

The persona defines "who it is," and the response rules define "how it will behave." This is where you convey your reservation and guest-welcoming discipline to the AI. With the AI Manager Chat, you add rules by writing in plain language, just as you would instruct a teammate.

Typical rules for hotels:

• "When asked about availability or price, first ask the check-in/check-out dates and party size (adults/children); then check the reservation engine."
• "If you’re unsure about price or availability, don’t make it up; say you don’t know and hand off to the team."
• "When asked about cancellation and deposit, read the policy from the knowledge base clearly; add no commentary."
• "When asked about child age, state the age-based pricing rule clearly; don’t guess."
• "On a group/event request (10+ rooms), direct them to the team for a custom offer."

The most common mistake in accommodation is the AI quoting a price or availability it is unsure of just "to say something," and that turning into a double booking, a cancellation, or a bad review. In inTusell, this boundary is two-layered. The primary control is the hotel-specific instructions given to the AI: from the start, the AI is constrained so it won’t step outside the knowledge base and engine and invent price/availability. On top of that, an output filter (a pattern scan plus a second-model check) scans replies for misleading health and legal promises and flags them; it does not block everything. In other words, the system’s reliability comes not from "a wall that blocks everything," but from the AI being correctly constrained from the start and from the accuracy of your room/price information.

To compare two AI tones, you can use the A/B test, which is in beta: for example, a more formal and a friendlier tone run side by side, each on half of the replies, and the experiment card shows which one leads to more wins. Today only the tone changes; the content and the rules stay the same.

5. Teach from past sales conversations

This is the step that moves training from "good" to "specific to your property." In the panel, you upload audio recordings of your past reservation conversations (MP3, MP4, WAV, M4A) and mark each one as Won or Lost. In accommodation, "won" means a conversation that turned into a reservation; "lost" means a guest who asked the price and left, or booked another property.

The system uses these recordings in two ways:

This way, the AI learns how your best reservations agent explained the advantage of booking direct to someone who said "it’s cheaper on Booking," and how they moved an undecided guest toward firming up dates, and applies the same approach in similar situations. Personal data is handled too: uploading requires explicit consent, and phone, email, ID and card numbers are masked in the transcript before processing.

This step is not mandatory, but don’t skip it. The knowledge base teaches the AI "what" it knows; past conversations teach "how you sell." When the two are combined, the assistant truly resembles an experienced reservations advisor.

6. Review and turn fixes into rules (ongoing step)

The first five steps get the AI operational; from there on you sharpen it for your property. You check replies in the AI Test screen by asking questions as a guest would, and in Conversations on real chats.

When you see a wrong or missing answer, make the fix permanent. Say the AI used the wrong name for a room type or left out part of the late check-out terms: add the correct information to the knowledge base. If it described a promotion wrongly or did not ask for dates before quoting, write a rule in AI Rules, switch on a fitting ready rule pack in the Behavior Library, or tell the AI Manager Chat in a plain sentence ("ask dates and party size before giving a price"). In a period when you want to see every reply before it goes out, put the account in Full approval mode or a single conversation in Approval mode.

The Label Manager is a separate job: a customer satisfaction reply, the outcome of a handed-off conversation, a won/lost classification and a label a team member adds by hand land there. You approve, reject or edit; corrections are kept as a quality record, but they do not change how the AI writes replies today. What changes behavior is the knowledge base and the rules. We recommend a short daily look in the first weeks.

7. Open the channels (ongoing step)

Once training is ready, you put the assistant in front of guests. inTusell gathers all channels in a single inbox: WhatsApp, Instagram DM, Instagram comments, Telegram, and Web Chat. Whatever channel the guest writes from, the same trained assistant answers with the same information.

Channel activation methods:

• Instagram: connects through Meta login; you pick your Facebook Page and Instagram account, with no keys or tokens to copy. The account must be a professional account linked to a Facebook Page.
• WhatsApp: under My business › Channels › Messaging Channels you press "QR ile Bağlan" (Connect with QR) and scan the code on your phone via WhatsApp › Settings › Linked Devices › Link a Device; if you have no second screen, "Link with phone number" gives you an 8-character linking code. The number stays the same and no Meta Business approval is needed. The official WhatsApp line (Cloud API) opens after Meta approval; today you connect with QR or a linking code.

As a first step we recommend opening WhatsApp and Instagram DM; in accommodation these two carry most of the incoming booking traffic. Once Instagram is connected, DM replies are already on. Comment replies have their own switch on the Instagram card ("Instagram comments"), which is on by default; if you want to answer "price?" and "available?" comments under room posts by hand for a while, you can turn it off. If you connected Instagram before September 24, 2026, you need to reconnect it once for comments.

Account-wide there are three AI modes. Intern (default) sends replies itself and asks you before a cancellation or refund. Senior also handles cancellations and refunds itself. In Full approval every reply on WhatsApp, Instagram and Telegram waits for your approval in Decisions › Approval Queue. Whatever the mode, when the AI is unsure or a case needs a person, such as a group request, it hands the conversation to your team; the handoff lands on the Decisions screen with the full history. A file, image, audio or video attachment is not by itself a reason to hand off; the AI tries to make sense of it and hands off only when it cannot complete the task.

Once the channels are open, CRM and proactive follow-up come into play: a guest who says "I got the price, let me think" and disappears gets at most 3 follow-up messages written by the AI; they go out only within business hours on WhatsApp, Instagram or Telegram and stop when the guest replies or asks not to be contacted. The full detail of daily operations (from an incoming message to a booking, payment and handing off to the team) is in a separate article: how a hotel uses inTusell. For channel setup details, the Instagram and WhatsApp sales automation article is ready too.

How long does training take?

How long it takes depends mostly on whether your documents are ready; if your room information is ready, the first upload is short. The steps of a working setup:

1. Uploading room types and the first files to the knowledge base
2. Defining room types, rate plans and availability in the Stays module
3. Persona, tone, and first response rules
4. Connecting the channels
5. First test conversations

But there is no moment of "training is done." In the first weeks, as you add what you notice in AI Test and in real conversations to the knowledge base and the AI rules, the assistant sharpens to fit your property; past call recordings and A/B tests are ongoing improvements too. Setup is done once; mastery is continuous.

What it isn’t

Putting inTusell in the right category matters, because the wrong expectation leads to the wrong setup.

• It is not a chatbot that gives canned answers. It is not a decision tree but an AI sales assistant and CRM that answers messages on WhatsApp, Instagram and web chat, opens reservations and follows each guest through the sales pipeline.
• It is not a tool that fabricates price/availability. It takes general information from the knowledge base and price/availability from the Stays engine; it says what it is unsure about and hands off to the team.
• It is not a channel manager or a PMS. It does not sync availability or rates with channels like Booking and Airbnb; it does not replace your property. It runs guest communication and direct bookings on your direct channels (WhatsApp, Instagram, web).
• It is not a service that announces open rooms. It is not a system that sends a bulk notification when a room opens up; it is an assistant that sends a guest who went quiet at most 3 follow-up messages within business hours.

In short: not a bot that spits out an answer to a single question, but a reservation and sales assistant that represents your property to the extent you train it and does not invent what it doesn’t know.

After training: reservations, payment, and proactive follow-up

After you train the assistant, two modules come into play. The first is the room reservation engine (Stays module, beta): you define your room types, capacities and date-range rate plans. The AI then asks for dates and party size on WhatsApp or Instagram, pulls the price for those dates from the engine, and creates a reservation by allocating the room in one locked transaction. For payment, bank-transfer details (amount, IBAN and reference code in one message) or an iyzico payment link go out; you approve bank-transfer receipts. Unlike appointments, there is no automatic reminder before a stay; you record cancellations and no-shows in the panel. Details: hotel room reservation automation.

The second is proactive follow-up (CRM): a guest who asked a price and did not decide, or left a conversation half-finished, gets at most 3 follow-up messages written by the AI, only within business hours, on WhatsApp, Instagram or Telegram. You will find how to run the assistant you trained in daily operations in the usage article of the series: how a hotel uses inTusell.

Frequently asked questions

How long does hotel AI training take?

A working setup comes down to these steps: uploading room types, prices, and policies to the knowledge base, setting the persona and tone, a few response rules, and connecting channels. How long it takes depends mostly on how ready your documents are; the real refinement comes in the first weeks, as you add what you notice in AI Test and in real conversations to the knowledge base and the AI rules. Training is not a one-time event; it is continuous.

Do I need technical knowledge to train the AI?

No. All training is done from the panel; no code, API key, or developer is required. You upload room and price information to the knowledge base as PDF/Excel/CSV, have your website scanned, choose the persona, and write response rules in plain language. You connect Instagram through Meta login and WhatsApp with a QR code or a linking code.

Will the AI give a wrong price or mention a room that does not exist?

No. The AI gives general information only from the room, price, and policy information you upload to the knowledge base; with the Stays module on, it pulls real availability and price from the reservation engine. It does not invent a nightly rate it is unsure of or a room type that does not exist; it says it does not know and hands off to the team.

How does the AI know real availability and price?

General information (room types, amenities, policies) comes from the knowledge base. Date-specific availability and nightly rate come from the room reservation engine when the Stays module (beta) is on: the AI asks for dates and party size, the engine calculates the price from the rate plan and allocates the room in one locked transaction; within inTusell the same room is not sold twice. With the module off, the AI makes no availability promise.

Which channels can I connect?

WhatsApp, Instagram DM, Instagram comments, Telegram, and web chat are gathered in a single inbox. Instagram connects through Meta login; WhatsApp links with a QR code or a linking code sent for your phone number. Comment replies are on by default; if you connected Instagram before September 24, 2026, you need to reconnect it once for comments.

How is my data protected under KVKK?

Data is stored encrypted and each business’s data is kept separate. You download your account data with "Download my data (KVKK)" and manage the KVKK notice your customers receive, with your own privacy link, under "Legal notices". Uploading call recordings requires explicit consent and personal numbers are masked in the transcript; guest ID and passport numbers are stored encrypted and the list shows only the last four digits.

Next step

You’ve trained your assistant; next comes using it in daily operations. The next article in the series explains how to run the inTusell you trained on a real reservation day: incoming availability questions, the reservation flow, payment, cancellations, and handing off to the team. Continue straight from there: how a hotel uses inTusell. For the details of the availability and price engine, the hotel room reservation automation article and, for channel setup, the Instagram and WhatsApp sales automation article are also ready.

This guide was written for hotels and accommodation, but the same 5-step logic works in every sector. You can also take a look at the main articles for neighboring sectors: you’ll find the same setup in e-commerce AI training for e-commerce, tour agency AI training for tourism, and clinic AI training for health. If you’d like to see a setup tailored to your property live, use Request a demo and we’ll open your panel together in a short session, or write to hi@intusell.com. You can find the full set of hotel capabilities on the hotel solution and solutions pages, package and quota details on the pricing page, and other guides in the all articles list.