For Jonathan Chan, Head of Global Operational Excellence and Innovation at Julius Baer, hackathons hold a special allure. “I always love seeing the solutions that emerge from these events,” he says. “It’s phenomenal to see what participants can achieve in 48 hours with a couple of boxes of pizza and a computer.” Having spent his career at the forefront of corporate innovation, Jonathan has repeatedly seen the power of hackathons in sparking new thinking and driving innovation. “Hackathons take us out of our daily routine to focus on a single problem,” he says. “You never know what their results will be, but one thing is certain – they keep us on our toes and inspire us to think outside the box.”
Tobias Amiet, Head of Products and Services Compliance in the bank’s Chief Risk Office (CRO), shares this view. As chair of the CRO Innovation Board, a bank-internal platform to facilitate bottom-up innovation within CRO and beyond, he uses hackathons to explore new approaches that address pertinent questions in risk management and compliance. “Hackathons demonstrate the potential of co-creation,” he says. “By bringing together diverse perspectives and leveraging the latest technologies, they show us what’s possible.”
In a cross-functional collaborative effort that involved both Jonathan’s and Tobias’ teams as well as IT and Private Banking Compliance, the bank set a challenge for hackers at two prominent events in Switzerland: SwissHacks, the Swiss financial industry’s largest focusing on fintech and regtech, organised by the Swiss Financial Innovation Desk (FIND), and startup incubator Tenity; and Datathon, the flagship event of the Analytics Club at ETH Zurich, which attracts top computer science and data science students from across Europe.
Empowering the client onboarding with AI
Last year at SwissHacks, Julius Baer’s challenge focussed on detecting deepfakes, an emerging threat that banks must safeguard against to protect clients and their wealth. This time, the bank chose a topic at the heart of private banking: improving the client onboarding process with artificial intelligence.
“In private banking, onboarding a new client lays the foundation for our relationship with the client and ensures we’re meeting our regulatory obligations while building trust,” explains Catherine Scholler, Head of Private Banking Compliance, and as a jury member an important contributor to the challenge. Moreover, the data points, attributes, and elements captured in the process steer a range of other tasks. “The quality of the data and how well the various stakeholders work together throughout the onboarding ultimately defines the efficiency of the process, its effectiveness, and the client service experience,” according to Catherine Scholler. This is where AI comes in by acting as a powerful assistant – automating routine tasks, improving data analysis, increasing operational efficiency, and enhancing risk management. As Tobias observes, “We saw an opportunity for AI and got excited about testing its potential at the hackathons.”
A challenge with a twist
In the resulting challenge presented at Datathon and SwissHacks participants were tasked with developing and training an AI model to accurately determine whether a client’s onboarding data met the necessary regulatory requirements. In the 48-hour SwissHacks event, the Julius Baer team introduced a twist to the task: instead of receiving the dataset as usual, participants worked with data integrated into a game.
“The game we devised – the ‘Client Onboarding Odyssey’ – is a simulation of client portfolios containing synthetic onboarding data,” says Michael Zemp, Head of Service Engineering at Julius Baer. “In the game, the player takes on the role of the person responsible for the client onboarding and has to assess whether the data complies with the rules.” He further explains, “However, the hackathon participants weren’t given the specific set of rules for accepting or rejecting a client, and they weren’t expected to play the game themselves.”
Their task was to design an AI that could learn to navigate the game. “The key difficulty for participants was enabling the AI to discover the underlying rules which they could then use to automate the process further,” he notes.
Michael Zemp and his team of experienced AI and machine learning (ML) engineers and graduates from the bank’s graduate programme were instrumental in conceptualising the idea behind the challenge and spearheading its technical implementation. “In my view, this is the most sophisticated challenge we’ve ever created – one that’s rooted in our core business and presented in an engaging way,” he says.
Breaking new ground
The opportunity to tackle a challenge that is both business-relevant and engaging also resonated with the hackathons’ participants. At Datathon, where the challenge had been tailored to fit the event’s condensed 24-hour format, the winning team, ‘Pepas’, convinced the jury with their model’s high accuracy of nearly 94 per cent in data validation, as well as demonstrating meticulous attention to detail and good coding standard.
At SwissHacks, with 220 participants overall and twelve teams working on the Julius Baer challenge, the winning team, ‘3plus1’ swept the competition, taking home not only the Julius Baer challenge win but also the overall SwissHacks prize and the audience prize. Harnessing the power of AI tools, their solution, ‘Smart data validation’, not only achieved the highest score in the client onboarding game, but also included a user-friendly application with a range of helpful features to effectively identify issues in the data.
More than a weekend of coding
However, the true measure of success lies not merely in the winning solutions, but in the potential for real-life impact. And this is where it gets really interesting for the winning teams who will meet with Julius Baer’s technology leaders to further discuss their solutions and possibilities to collaborate. As Tobias Amiet notes, “The outcome from the hackathons clearly exceeded my expectations and we’re excited to explore how we can apply these innovative ideas in the bank.”
At Julius Baer, embracing the potential of AI to drive digital business transformation and deliver technology-enhanced client services and experiences is a key priority and pushed broadly across the organisation.
With his finger on the pulse of corporate innovation, Jonathan Chan sees the impact of events such as SwissHacks in an even bigger context. “We are in midst of an innovation super cycle,” he comments, with reference to Julius Baer’s Secular Outlook, and adds, “AI will continue to reshape industries and its potential in wealth management is vast. At Julius Baer, our open and collaborative innovation approach positions us to keep our edge and play our part in this evolution.”