Radiotelevisione Svizzera (RSI), Swiss Broadcasting Corporation’s Italian-speaking business unit, is developing a new all-in-one mobile app. This venture, powered by artificial intelligence (AI) and machine learning (ML) technologies on AWS, caters to users’ demand for more personalized and dynamic media experiences.
Swiss Broadcasting Corporation is a Swiss public broadcasting organization and one of the largest media houses in Switzerland. Serving Italian speakers, its RSI business unit delivered news, sports, cultural, and entertainment content through three disparate apps and a website. To enhance the user experience, the company wanted to combine its digital offerings into a single app so that viewers could access their desired content when and where they wanted it.
“Our idea was to develop an all-in-one application that can collect all of the content that we generate every day,” says Massimiliano Babbucci, head of digital innovation at RSI. “To accomplish this efficiently, we wanted to incorporate AI/ML to automate several parts of our workflow, from collecting metadata to recommending content and personalizing the user’s feed.”
To bring this vision to life, RSI turned to
Claranet Switzerland (Claranet), an
AWS Premier Partner that offers adaptive, scalable technology solutions in the cloud. Having worked with Claranet since 2019, RSI knew that the company held deep expertise in AWS services that it could draw upon to build the new app. When it came to building a large language model using AWS AI/ML technology, however, the project required additional expertise. To deepen and expand this collaboration, RSI also began to work with researchers and students at the Institute of Information Systems and Networking (ISIN), a research institute within the Department of Innovative Technologies at the
University of Applied Sciences and Arts of Southern Switzerland (SUPSI), which is based in Lugano.
“We established a technological hub between RSI, Claranet, and the university,” says Babbucci. “In a very short time, we were able to begin to develop the app with experienced data scientists without the need to train new people. ISIN is in charge of research and defining the best models to use, and then it provides the prototypes to Claranet to standardize and put into production.”
The RSI mobile app is a large language model that incorporates AI/ML technologies that help extract metadata and recommend relevant information to users. For example, ISIN adopted Amazon Comprehend—a natural-language processing service that uses ML to uncover valuable insights and connections in text—to extract relevant language, classify content, and personalize user feeds. Many of these models run on Amazon SageMaker, a service that is used to build, train, and deploy ML models for any use case.
“ISIN researchers first identified some viable models that they can use out of the box,” says Piero Bozzolo, AWS solutions architect at Claranet. “Then, they developed some models from scratch and discovered how to integrate them with Amazon SageMaker.”
The app is hosted on managed AWS serverless services, including
AWS Lambda, a serverless, event-driven compute service. RSI also has the option to use
Amazon SageMaker Serverless Inference—a purpose-built inference option that lets developers deploy and scale ML models without configuring or managing the underlying infrastructure—as needed.
“Currently, 90–95 percent of the infrastructure is serverless,” says Bozzolo. “One thing that I appreciate about Amazon SageMaker is that you can change the whole infrastructure base; you can switch from serverless and provision without much effort.”
RSI has realized several advantages by using this flexible infrastructure. “Using a serverless-first approach on AWS, RSI can scale up to reach millions of customers at any time,” says Bozzolo. “They can do this without worrying about the cost or performance of the application; they can maintain the same level of performance whether they have two concurrent users or 2 million.” Moreover, because RSI’s development team no longer needs to worry about hardware maintenance, they can focus on services and innovation rather than on low-level technical details.
The people involved in this project have enjoyed many benefits as well. Students at the Department of Innovative Technologies of SUPSI had the opportunity to gain practical experience by working on this real-world project. This hands-on experience, including meeting production deadlines and dealing with real-world challenges, provides valuable learning experiences that go beyond traditional university projects. Additionally, RSI journalists benefit from better tools and improved searchability. The automated metadata extraction engine has enhanced the quality of its tagging systems, which helps journalists quickly search for and find the content that they need when they need it.
The development of the RSI app is still ongoing, and several key features remain on the company’s road map. Over the next 2 years, RSI will focus on adding these new elements and turning the concept of omnichannel engagement into a reality. The collaboration between RSI, Claranet, and ISIN will serve as a robust foundation for these future enhancements.
“As a public service broadcaster, when you have to do a project, especially if it’s a really complex one like this, then you have to be really close to the people that you trust with it,” says Babbucci. “This is why I prefer to use AWS. I never had any doubt about using AWS.”
