Guided walkthrough
To get the most out of INTO AI, it's important to understand how our Message engine and Reasoning engine collaborate to craft the perfect responses to guest inquiries.
As illustrated in the diagram below, whenever you receive an inquiry or message from a guest, the information undergoes a series of sophisticated processes that result in instant, accurate responses in any language, 24/7. We achieve this by leveraging cutting-edge Large Language Models (LLMs), powerful knowledge classification and retrieval systems, and proprietary techniques. These ensure that our message engine generates highly accurate responses that directly address the guest's questions, sticking strictly to the provided information without making assumptions or inventing details.
- Understanding and Labeling: Our Reasoning engine first understands the context of the inquiry and labels it accordingly.
- Orchestration: With the inquiry labeled, the reasoning engine passes the information to our orchestrator, which determines the best course of action. This might involve searching the knowledge base for relevant information or calling a third-party service to perform tasks like making recommendations or booking a restaurant.
- Confidence Scoring: Once the response is generated, the reasoning engine applies a confidence score to determine the next steps.
Response Delivery:
Based on the confidence score, the response is either sent directly to the guest if INTO Autopilot is enabled or forwarded to an agent through INTO Copilot in your PMS inbox.
Read Next
- Make sure your data is accurate and up to date