The agent allows users (e.g., hotel reception staff) to ask natural language questions about one or more reservations and, depending on their question, receive information about these reservations (duration, room type, board basis, etc.) or suggestions tailored to the customer's profile that can help increase sales value.

The database contains tables with reservation history, including information on room type, number of nights, number of people, and other relevant reservation data.

Procedure documents are texts stored in the database itself or in OCI Object Storage that contain instructions defining which complementary sales products are applicable according to the reservation characteristics, the customer, and the establishment's policies: recommending breakfast for room-only reservations, proposing a room upgrade, etc.

Oracle's Select AI is a tool that allows users to interact with databases using natural language. This technology, integrated into Oracle Autonomous Database, facilitates the generation of SQL queries from natural language questions, using large language models (LLMs) and retrieval-augmented generation (RAG) techniques.

The agent's response elaboration is executed in several phases:

1. Based on the received query, it is first determined whether the entered question needs additional context, such as information previously provided in the conversation.
2. Next, the question's intent is determined. In the use case, for example, it is identified whether the question refers to one or more reservations or if recommendations on hotel services or city attractions are requested.
3. Once the intent is determined, the question is rewritten according to it, adding the necessary context and a specific prompt for each intent and context.

a. If the question refers only to reservation information, using Select AI, a query about the reservations is automatically generated and executed on the database, and the obtained data along with the context is sent to the generative AI to elaborate and deliver the information about those reservations in natural language.

b. If the question refers to additional or promotional sales recommendations for a specific reservation, a query about the reservation is also generated with Select AI, executed on the database, and the returned results along with the context and, if necessary, the procedure document with complementary sales instructions are sent to the generative AI. With all this information, the AI elaborates the response, and the user will receive the reservation details along with a list of additional sales products or promotions that best fit the customer's characteristics.

Using the agent significantly improves the quality of the customer offer by providing personalized recommendations and upselling options during the check-in process that would otherwise be slow to check, require authorization, or might not be suitable for their profile. This not only facilitates an increase in sales but also enhances guest satisfaction by offering services and promotions tailored to their needs and preferences.

Although the agent presented in this case has been specifically created for the hotel industry, its internal design allows it to be adapted to any other industry. The key lies in the combination of the AI capabilities integrated into Oracle Autonomous 23ai and the techniques developed by Quistor to obtain information by integrating the results of queries on data in tables and procedures in natural language. From here, the agent can be adjusted to the particularities of each business.

The ability to generate endpoints from the database objects on which the agent's logic is based facilitates integration with other applications, offering a flexible and adaptable solution to the specific needs of each industry. This flexibility and adaptability make the AI agent usable in sectors such as retail, insurance, advertising, banking, and many others where personalization and improvement of the customer experience are equally important.

Moreover, Oracle technology ensures that data is managed efficiently and securely, allowing companies to quickly access relevant information and respond accurately to customer queries and needs.

The implementation of this solution demonstrates the potential of Oracle's artificial intelligence to transform customer interaction and optimize operational processes in various sectors. At Quistor, thanks to our knowledge and experience, we can leverage this potential to offer a powerful and versatile tool that not only improves information retrieval efficiency and customer satisfaction but also enhances sales quality.

In summary, the use of the AI agent developed by Quistor, built on Oracle technology, provides the industry with a powerful tool to improve customer service quality, increase sales, and stand out in an increasingly competitive market.