Context

In March 2020, the AI research institute Mila approached Libéo: the plan was to set up a team and develop an advanced technology tool. A team of over 20 digital experts was on it a few days later.

A major effort began to create a leading-edge AI-enabled app with the primary aim of reducing the spread of the virus throughout the country. The technology is based on a peer-to-peer contact tracing system thought up by a coalition of researchers and experts led by Yoshua Bengio from Mila.

After several weeks of international collaboration with researchers from the University of Toronto, Stanford University and MIT, the COVI app was born. The app was capable of predicting the risk of COVID-19 infection. The mobile technology assesses the user’s likelihood of infection and anonymously shares the information with his or her contacts so that they can evaluate their own level of risk, as well as the risk of spreading it to their contacts, etc.

Thanks to our extensive experience in software development and our multidisciplinary team, we were able to create a world-class application within a tight timeframe. We were also able to achieve this feat due to our unique user-centred approach and rigorous development practices.

Technologies

• Artificial Intelligence (AI) – developed by MILA
• Flutter (SDK user interface to build apps from a single codebase)
• DNS infrastructure for message exchange
• Bluetooth communication

Our solution 

An app that meets the highest standards of data privacy and protection

You can’t build a contact tracing app without making privacy a top priority.

That’s why Libéo and Mila went out of their way to comply with all Canadian laws. The COVI app complies with all the principles set out in the Joint Statement by Federal, Provincial and Territorial Privacy Commissioners. It also complies with the Montreal Declaration on Responsible AI and is supported by UNESCO.

For transparency purposes, the non-profit organization COVI Canada took charge of managing the data collected to prevent any secondary use, such as use for commercial purposes.

COVI’s leading-edge technology was also based on a healthy philosophy of mutual aid and learning. The project was always guided by important ethical principles.

A leading-edge app

COVI is more than just a contact tracing app—it’s enabled by advanced artificial intelligence so that it can deliver a higher level of predictability.

The AI considers multiple factors, such as gender, age and symptoms, to estimate a personal level of risk. The goal was to outperform traditional methods and be able to predict the risk of contagion in specific areas.

Working on COVI during the pandemic

COVI is a major achievement for Libéo and presented multiple logistical challenges that we were able to overcome. Of course, the particular context of working remotely during a global pandemic contributed to its unique challenges.

To complete such a complex project in a tight timeframe, Libéo used the AGILE development methodology with two-week cycles. That enabled the team to make adjustments to the project based on new government directives or issues that arose along the way.

Another one for the books

The COVI app is a great example of the exceptional work our team is able to produce in a short period of time. In the end, the app was not selected by the Canadian government, which preferred a less innovative technology developed by Apple and Google.

COVI uses AI to predict future waves of COVID-19. It’s a useful app with an educational purpose. As we know how useful and valuable such detailed and thorough work can be to others, we shared COVI’s source code under an open-source license.