Adding New Knowledge
Add knowledge to improve AI response quality
Overview
The AI is only as good as the information it has, so improving response quality requires continuously adding new and relevant knowledge. In HostAI, knowledge can be broken down into its most basic elements, called blocks. A block of knowledge can be something as simple as a Wi-Fi password for a particular listing, food recommendations for a group of listings in the same city, or the company’s Airbnb cancellation policy.
Knowledge can apply to one listing, multiple listings, or globally across all listings.
Workflow
When adding new information to your knowledge base, it’s essential to determine the scope of the information. Ask yourself: "Does this knowledge apply to one specific listing, a group of listings, or all of them?"
Knowledge in HostAI can be added on a:
Listing-specific level – only applies to one listing
Multi-listing level – applies to a group of listings
Global level – applies to all listings
It’s also important to distinguish between information and rules. While information includes factual data you want the AI to know (e.g., cancellation policies, amenities), rules provide instructions for how the AI should behave in specific scenarios (e.g., how to handle early check-in requests).
Steps to Follow
Determine the Scope: Decide whether the information you're adding is listing-specific, multi-listing, or global.
Add New Knowledge:
Go to the listings page to locate where the new information should be applied.
You can add blocks of knowledge related to specific listings (e.g., listing amenities, Wi-Fi password), groups of listings (e.g., local food recommendations), or globally in Global Settings (e.g., company-wide refund policy).
Rules for Behavior: If you want to control how the AI behaves in specific situations, you can add rules in Global Settings alongside Global Info. Rules act as guardrails to guide the AI's responses, ensuring it behaves appropriately for each scenario.
Adding Rules for AI Behavior
Rules in HostAI are essential for guiding how the AI responds in specific scenarios, helping you control the AI's actions to ensure it meets your desired communication style. Rules can be created based on various variables like reservation status, dates, or the guest's message. When defining rules, it’s important to include these variables so the AI can interpret and follow them accordingly.
Key Variables to Use in Rules
Here are some commonly used variables that can be incorporated into your rules to tailor the AI’s behavior:
Natural Language Instruction | Variable to Use |
Reservation status | "reservation status" |
Date of the guest message | "current listing date and time" |
Check-in date | "check-in date and time" |
Check-out date | "check-out date and time" |
Stay length | "number of nights" |
Booking channel | "booking channel" |
Guest name | "guest name" |
There is a same-day turnover | "requested dates are unavailable" |
There is no same-day turnover | "requested dates are available" |
Example Rules
1. Early Check-In Request > 48 Hours Before Check-In
Objective: Prevent approval of early check-ins more than 48 hours before check-in and inform the guest to check back closer to the check-in date.
Rule:
2. Late Checkout Request Within 24 Hours of Check-Out
Objective: Handle late checkout requests within 24 hours before the check-out date by providing a personalized response.
Rule:
3. Cancellation Policy Based on Booking Channel
Objective: Deliver the correct cancellation policy based on the guest's booking channel (Airbnb, direct booking, etc.).
Rule:
4. Special Offer for Returning Guests
Objective: Offer a special deal to guests who had a positive previous stay.
Rule:
Last updated