Banking on the future: RBC’s Borealis and Digital teams collaborate to help Canadians better manage their finances
Read Time : 3 minutes
Artificial intelligence is changing nearly every sector of society, powering new types of products and services that are now common in daily life. Since the start of the pandemic, we’ve become increasingly reliant on the technology, whether it’s helping to recommend an online show or clothing purchase, or streamlining how we manage our finances.
AI technology has been a part of RBC for more than thirty years, providing important insights and security measures for clients, while driving efficiencies in business operations. And now, with the launch of NOMI Forecast, it is taking centre stage in our mobile banking platform.
In 2021, Borealis AI, the world-class AI research centre backed by RBC, was tasked with building an AI-capable feature to provide clients with a quick view into their future cashflow. The goal was to reimagine the role we play in the lives of RBC’s customers, delivering to them personalized insights and advice to help build confidence in managing their money.
“Cashflow forecasting is a notoriously challenging problem to solve, but through our strong collaboration with RBC’s Digital team we were able to deliver next generation cashflow forecasting in the palm of our clients’ hands,” said Alex LaPlante, Senior Director, Product and Business Development, Borealis AI. “One of the early challenges we solved together was where to start. We chose to tackle pre-authorized payments first, as it makes up approximately 50% of all transactions in a chequing account.”
And so, the Borealis AI team began working closely with RBC’s Digital team to build a new financial capability that’s quickly become a game-changer for digital clients.
Bringing together two teams, working in three different cities, with very different skill sets was not without its challenges. However, the combination of machine learning and deep digital money management experience created the “secret sauce” for success, and in fact, has become a collaboration model for the enterprise.
“Being a part of the team that rallied around this new capability was exhilarating,” said Maryam Khoshyomn, Director, Digital Money Management, RBC and Product Lead for NOMI. “We had AI researchers developing unique algorithms, software engineers deploying them into production, and our digital team leveraging the best practices and learnings from RBC’s other award-winning NOMI capabilities. The range and depth of expertise helped supercharge the launch of NOMI Forecast.”
In late 2021, NOMI Forecast was launched to provide clients with a quick seven-day forecast into their future cashflow by tracking recurring bill payments, giving them a more holistic view of their account activity. It is the first product to integrate Borealis AI’s technology into RBC’s retail banking space, building off the success of Aiden, an award-winning AI-based electronic trading platform that uses the computational power of deep reinforcement learning in its pursuit of improved trading results and insights for clients.
NOMI Forecast was built on Borealis AI’s OpenShift GPU cluster, which is optimized for building large scale machine learning models. This enabled the team to develop an algorithm that can predict upcoming payment dates and amounts for multiple different payment types including pre-authorized payments, money transfers, cheques and e-transfers.
“Our goal at Borealis AI is to remove the complexities of building and running machine learning models for our partners, in this case for RBC’s Digital team, to help them focus on how they can evolve their products and systems to incorporate AI capabilities,” said Payam Sabzmeydani, Senior Machine Learning Engineering Lead at Borealis AI.
“Beyond building the model, we also take on the ownership and responsibilities for its production operation using our in-house tools and expertise. This is usually done by decoupling the AI/ML systems through application interfaces that outline the responsibilities of each group in well-defined agreements to allow each team to incorporate their expertise in building a unified product for RBC.”
The Borealis AI-built machine learning model behind NOMI Forecast uses deep learning and historical transaction data to forecast future transactions at a high level of accuracy. Since launching, NOMI Forecast has served over 1.7 million predictions to RBC clients.
“The launch of NOMI Forecast is a great example of how we continue to harness the power of data-driven insights by bringing the right people together and leveraging our different expertise, skills and backgrounds,” says Khoshyomn.
Borealis AI and our digital business have plans to collaborate further to create additional value for RBC’s clients, with plans to accelerate and adopt new technologies that outpace expectations.
Photo Caption: The Borealis AI team that worked on NOMI Forecasting: Ella Wilson, Ryan Kaden, Yongmin Yan, Mahyar Hosseini, Yunduz Rakhmangulova, Amir Abdi, Lili Meng, Tristan Sylvain, Gabriel Oliveira, Payam Sabzmeydani
