Case Study

UnipolSai renews one-to-one interaction with customers through a new digital assistant

Through the collaboration with Machine Learning Reply, UnipolSai launches a chatbot that uses generative artificial intelligence to provide a more engaging and personalized service to its customers.

#Generative AI
#Digital Assistant
#Customer Experience



Create a digital assistant able to guide customers in choosing the service that best suits their needs


Towards a digital communication that is more attentive to needs

UnipolSai, one of the main insurance companies in Italy, with a wide portfolio of business and consumer products and services, had started a project to evolve its consumer offer for the home, with the aim of making it more flexible and focused on customer needs.

In this context of transformation, UnipolSai has recognised the importance of also redefining its digital communication strategy and of evolving the chatbot dedicated to managing product information requests towards a generative conversational model, so as to be able to respond in a more relevant and complete way to customer requests and needs.


A new digital assistant for quality customer service

The result was a digital assistant powered by generative artificial intelligence, capable of managing conversations in a similar way to human conversations about UnipolSai's home offer. The central element of the solution is the personalisation of interactions, which makes it possible to provide quick and relevant answers while reducing support requests to service centers.

Furthermore, the chatbot has been designed to integrate harmoniously with UnipolSai's existing infrastructure, and to be easily adaptable to other insurance products and services of the company.



The RAG approach for precise and contextual answers

UnipolSai's digital assistant is based on a hybrid architecture that combines on-premise and cloud solutions, making use of various services: the chatbot's backend uses Google Dialogflow to manage basic interactions, while the generative component uses Azure Open AI services. Machine Learning Reply worked, first, on the preparation of the knowledge base, to make the insurance policy information documents easily usable by the generative model, and then on the specialisation of the Generative AI model.

With the aim of having a solution that is as precise and reliable as possible in the answers, the RAG (Retrieval Augmented Generation) approach was chosen, which allows the chatbot to select only the most relevant information within the knowledge base and use it to generate answers, and specific security mechanisms that have been implemented to limit hallucinations.

The benefits

Improved operational efficiency

This project marked a first step in the use of Generative AI in UnipolSai (in production since July 2023), as well as interesting potential benefits in operational efficiency and customer experience.


An interest was found in requests for quotes, indicative of an improvement in the interaction phase and a reduction in the abandonment rate.

Requests for contact
with Unipol staff

Customers who interface with the chatbot also require personalised advice with Unipolsai agents, to further deepen the details of the policy to be taken out.

Improving the
customer experience

The positive feedback received from customers regarding the effectiveness and personalisation of interactions with the chatbot, reflect a general increase in their satisfaction with the services offered.


UnipolSai Assicurazioni S.p.A. is the insurance company of the Unipol Group, a leader in Italy in the Non-Life sectors, in particular in the Auto and Health sectors. Also active in the Life branches, UnipolSai has a portfolio of more than 10 million customers and occupies a leading position in the national ranking of insurance groups for direct collection equal to 15.1 billion euros, of which 8.7 billion in the Non-Life Branches and 6.4 billion in the Life Branches (2023 data). The company operates through the largest agency network in Italy, with more than 2,300 insurance agencies distributed throughout the country. UnipolSai is also active in direct car insurance (Linear Assurances), transport and aviation insurance (SIAT), health protection (UniSalute), supplementary pension provision and oversees the bancassurance channel (Arca Vita and Arca Assurances). It also manages significant diversified activities in the real estate, hotel (UNA Group), medical-healthcare (Centro Medico Santagostino) and agriculture (Tenute del Cerro) sectors. UnipolSai Assicurazioni is controlled by Unipol Gruppo S.p.A. and, like the latter, is listed on the Italian Stock Exchange.


Machine Learning Reply is the Reply group company specialized in Machine Learning, Cognitive Computing and Artificial Intelligence solutions. Machine Learning Reply, based on the most recent developments in the field of artificial intelligence, applies innovative Generative AI, Deep Learning, Natural Language Processing, Image/Video Recognition techniques to different use scenarios such as Smart Automation, predictive engines, document processing, recommendation systems and conversational agents.