Skip to main content

Facts and FAQs: how the AI Agent learns

The two knowledge types the AI Agent reads — Facts and FAQs — and how they work together.

The AI Agent answers from a knowledge base you provide. There are two types of knowledge entries: Facts and FAQs. The quality of these directly determines the quality of the AI Agent's replies.

Facts

Facts are factual statements about your property. They are reference information the AI Agent can draw on when answering a guest.

Examples:

  • Check-in time is 15:00. Check-out time is 11:00.

  • The hotel address is 12 Hackney Road, London E2 7NS.

  • Breakfast is served from 07:00 to 10:30 in the ground-floor restaurant.

  • Parking is available at the public car park next door, £24 per 24 hours.

  • Wi-Fi network: HotelGuest. Password: welcome2024.

Facts are the building blocks. Many FAQs are answered by combining Facts.

FAQs

FAQs are explicit question-and-answer pairs. They handle the questions guests ask most often, in the form guests ask them.

Examples:

  • Q: What time is check-in? / A: Check-in is from 15:00. If you arrive earlier, we will hold your luggage at reception.

  • Q: Is there parking? / A: There is a public car park next door, £24 per 24 hours. We cannot reserve a space.

FAQs work best when they answer in the way the guest actually wants the answer, not just the factual minimum.

How the AI Agent uses them

When a guest message arrives, the AI Agent looks for relevant FAQs and Facts in its own knowledge section. If it finds a good match, it answers from that knowledge. If it cannot find a confident match, it hands off to a human.

What this means in practice

The AI Agent will only answer what you have put in. There is no general world knowledge it can fall back on for property-specific questions. A gap in the knowledge base is a gap in the AI Agent's capability.

See Writing effective FAQ entries for how to structure entries that the agent can actually use.

Need Help?

Please contact us through the 'Talk to Us' option on the left menu in the platform, or through the Bookboost Support email at support@bookboost.io if you have questions or need additional support.

Did this answer your question?