🤖 Chatbot Configuration

Explains how to personalize and optimize your hotel's AI agent behavior.

What is the ChatBook chatbot?

The chatbot is an AI sales agent that acts as your hotel's virtual receptionist, available 24/7. It learns about your hotel through a vector knowledge base (RAG) and adapts its responses to each guest's profile.

What the bot can do:

• Answer questions about the hotel, rooms, services, and location
• Show hotel photos when the guest requests them
• Quote prices and check availability (if you have a PMS integrated)
• Create direct bookings (Matrix/Pro with PMS)
• Capture leads with contact details
• Detect the guest type and adapt the conversation

What the bot does NOT do:

• Make up information — if it doesn't know, it asks for the guest's contact details
• Process payments directly (coming in Phase 3)
• Remember previous conversations beyond 10 messages

1. Agent personalization

Go to Hotels → [your hotel] → Settings in the side menu.

1.1 Bot identity

Field	Description	Example
Bot name	How it introduces itself to guests	"Sofía", "Hotel Luna Assistant"
Default language	The bot detects the guest's language automatically. This is the fallback	Spanish
Tone	Formal / Friendly (both are valid, choose based on the hotel's profile)	Friendly

1.2 Web widget configuration

Accessible from Connections → Web Widget → Customization:

• Primary color — color of the chat button and header
• Hotel logo — image that appears in the chat header
• Welcome message — the first message the guest sees when opening the chat
• Widget position — bottom-right corner (default) or bottom-left
• Text field placeholder — helper text in the chat input

1.3 Widget embed code example

<!-- Agregar antes de </body> en tu sitio web -->
<script>
  (function(d,s,id){
    var js,fjs=d.getElementsByTagName(s)[0];
    if(d.getElementById(id))return;
    js=d.createElement(s);js.id=id;
    js.src="https://widget.hotelchatbook.io/embed.js";
    js.setAttribute('data-hotel','TU_HOTEL_ID');
    fjs.parentNode.insertBefore(js,fjs);
  }(document,'script','chatbook-widget'));
</script>

2. Knowledge base — how the bot learns

The knowledge base is the bot's "brain". The more complete it is, the better it sells.

2.1 What the knowledge base includes

Category	Description	Source
General info	Name, description, address, phone, website	Initial scraping + manual
Rooms	Types, capacity, description, amenities	Rooms module
Services	Breakfast, pool, spa, parking, restaurant	Conversational training
Rates	Base prices by category and season	Rates module
Policies	Cancellation, check-in/out, pets, children	Hotel Information
Location	Exact address, Google Place ID, directions	Google Places
Images	Photos of rooms and common areas	Manual upload in AI Training
Custom FAQs	Answers to the hotel's frequently asked questions	Training + Missing Information

2.2 Knowledge quality

The bot responds with different levels of certainty:

• Confirmed certainty → answers directly with the data
• Partial certainty → answers with "Normally…" or "Subject to confirmation"
• No information → "I don't have confirmed information about this. Would you like the team to contact you?"

tip

To improve coverage, regularly review the "Missing Information" section in AI Training — that's where the questions the bot couldn't answer appear.

2.3 How the bot uses rates

The pricing engine follows this priority hierarchy:

1. Discount rules (if applicable — Matrix/Pro)
2. Traveler package price (if the guest picks one)
3. Active promotion with dates (% or fixed amount)
4. Seasonal rate (multiplier per date range)
5. Day-type rate (weekend / weekday)
6. Base rate (fallback)

3. Bot behavior by guest type (Buyer Personas)

The bot automatically detects the guest type and adapts its response. This happens in the bot's "internal monologue" — the guest doesn't see it.

Buyer Persona	Signals it detects	Bot strategy
BP1 — Direct Booker (35% of traffic)	Exact dates, direct price question, "I want to book"	Shows price immediately + booking CTA. Maximum conversion.
BP2 — Comparer (40% of traffic)	Compares options, tentative dates, "I'm looking around"	Shows unique value proposition + soft urgency + conditional price
BP3 — Explorer (20% of traffic)	"I'm thinking about…", no dates, exploratory tone	Inspires, does NOT show price. Captures email for nurturing
BP4 — Urgent (5% of traffic)	"Tonight", "right now", "urgent"	Immediate close. Availability first, then details

When the bot hands off to a human (Handoff)

The bot reports the lead and escalates to a human when:

1. The guest explicitly asks to speak with a human
2. Complex negotiation (large group, event)
3. 3+ messages with high intent without closing
4. Critical missing information that the bot cannot resolve

4. WhatsApp configuration (Matrix / Pro)

4.1 Connect a WhatsApp Business number

1. Go to Connections in the side menu
2. Select "WhatsApp Business"
3. Click "Connect number"
4. Options:
   • Your own number: you need an active WhatsApp Business API (via 360dialog or a similar provider)
   • ChatBook shared number: the hotel receives messages on a UNOZERO number (available in basic Matrix)
5. Enter the whatsapp_account_sid and auth_token from 360dialog
6. The system verifies the connection — it takes ~2 minutes
7. Expected status: connected

4.2 Message templates (Meta Templates)

WhatsApp requires Meta-approved templates to start conversations. The bot uses them for:

• Booking confirmation
• Payment reminder with deadline
• Post-registration welcome message

View and manage templates at: Connections → WhatsApp Business → Templates

info

Sending non-template messages within the 24h window (in reply to a guest's message) is permitted without restrictions.

5. Optimize bot performance

5.1 Quality checklist — a well-configured bot

Check	How to verify it
✅ Complete hotel info	AI Training → completeness % > 80%
✅ At least 5 images uploaded	AI Training → Images
✅ Check-in/out times configured	Hotels → [your hotel] → Policies
✅ Cancellation policy defined	Hotels → [your hotel] → Policies
✅ At least 1 room category with a rate	Hotels → [your hotel] → Rooms + Rates
✅ Missing information reviewed	AI Training → Missing Information (0 pending items)
✅ Widget visible and working on the website	Visit the website and start a test chat

5.2 Bot benchmark metrics

Metric	Current ChatBook value	Industry benchmark
Conversion rate (lead → booking)	18.4%	12–15%
Lead capture rate	64.2%	45–55%
Average response time	1.8 seconds	3–5 seconds
Guest CSAT	4.7 / 5.0	4.2 / 5.0
Escalation to human	8.3%	15–20%

5.3 Common performance issues

Symptom	Probable cause	Solution
Bot says "I don't have information" frequently	Incomplete knowledge base	Complete AI Training + review Missing Information
Bot shows incorrect prices	Rates not configured or PMS disconnected	Check the Rates module and PMS Integration
Bot doesn't show images	Images without consistent names	Rename using the {tipo}-{número} format
Bot doesn't capture leads	Notification channel not configured	Configure email or WhatsApp in Hotels → Notifications
Conversations fall into expired without a reply	30 conv/month limit reached (Freemium)	Buy an additional package or upgrade the plan