Case studies

AI Travel Assistant: a case study of La Filanda

Discover how La Filanda, a charming bed and breakfast on Lake Garda, is improving its booking process with cutting-edge AI technology by implementing an AI Travel Assistant. Dive into our latest case study to see how a Machine Learning model transforms how guest quotes are managed, enhancing efficiency and customer satisfaction. Watch the video of the final presentation of the project.

In the ever-evolving world of hospitality, bed and breakfast establishments are embracing the digital revolution to stay ahead. La Filanda, an agritourism located on Lake Garda, is at the forefront of this transformation. Known for its cosy ambience and in a historic village, La Filanda faces the challenge of managing dozens of daily booking requests.


The current system, which combines GPT-3.5 technology with hotel property management software, already shows promise. GPT-3.5 helps by extracting structured information from natural language booking requests. The hotel property management software then uses this data to create tailored quotes for each customer. However, despite its efficiency, this system is plagued by delays and resource-intensive processes, posing a risk to guest satisfaction and operational accuracy.

To address these challenges, La Filanda initiated an innovative project.

The goal? To streamline their booking process using a Machine Learning model. This model, trained on datasets provided by La Filanda, aims to mimic the expertise of a human operator in filtering and ranking quotes. The datasets include customer details, standard quotes, and options generated by the hotel property management software, among others.


The approach is straightforward yet sophisticated. We treat quote filtering as a classification problem. The Machine Learning model, particularly a Random Forest classifier, learns to predict the ‘goodness’ of a quotation, informed by human operators’ past decisions. It accelerates the quote generation process and ensures that only the most suitable options are presented to customers.


Our results are promising. With precision at 51% and an impressive recall of 88%, the model demonstrates its ability to identify quotes likely to be accepted by customers efficiently.
However, continuous re-training with new data is recommended to enhance its performance further.

In conclusion, La Filanda’s journey with AI in booking optimisation is a testament to the power of technology in transforming traditional industries. This case study showcases our commitment to innovation at Pi School and opens up new possibilities for the hospitality sector at large.

Do you have a similar challenge to solve? We can help you!