Estimated reading time: 6 minutes

In a collaborative effort to shape the future of the financial ecosystem, Sibos brings together thousands of business leaders, decision-makers and topic experts.

One panel discussion at this year’s event focussed on Swift’s recent artificial intelligence (AI) efforts with several of its key collaborators in the space who are working to advance responsible AI and data security practices.

The panel’s moderator, Tom Zschach, Chief Innovation Officer at Swift was joined by Kelly Switt, Global Head of Intelligent Edge Business Development at Red Hat; John Overton, CEO at Kove; Vivek Agarwal, Head of Financial Services Solutions at C3 AI; and Johan Bryssinck, AI Programme and Shared Services at Swift.

Together they explored the current landscape of financial crime prevention, laying the basis for why the space is ripe for technological innovation.

AI, explained

AI is the ability of machines to perform tasks that typically require human intelligence, such as learning, reasoning, and problem-solving. 

Generative AI is a type of AI that can create new content, such as text, images, music, and code, from scratch.

Enterprise-scale AI is the use of AI to solve complex business problems at scale. This means using AI to automate tasks, improve efficiency, and make better decisions across the entire organisation. Enterprise-scale AI requires a number of things, including:

  • AI platform: AI models are trained on data, so it is important to have the ability to integrate large amounts of data from a variety of internal and external sources. This data needs to be organised and fed into models relevant to the problem that you are trying to solve, and supported by guard rails to ensure responsible AI.
  • Computing power: AI models require a lot of computing power to train and deploy. Enterprise-scale AI often requires the use of cloud computing platforms or specialised AI hardware.
  • Expertise: AI is a complex field, and it is important to have expertise in AI development and deployment in order to implement enterprise-scale AI solutions.

Enterprise-scale AI is being used by a wide range of companies in a variety of industries. For example, banks are using AI to detect fraud and improve risk management. 

What is federated learning?

Federated learning is a machine learning technique that allows multiple devices to collaboratively train a single model without sharing their data.

Federated learning is useful for training machine learning models on sensitive data, such as medical records or financial data, without compromising the privacy of the data owners.

The imperative for change

Traditional approaches are becoming less effective in tackling evolving financial crimes, necessitating a shift towards more agile and intelligent systems. 

The new AI platform developed by Swift in collaboration with Red Hat, Kove, and C3 AI is designed to address this issue by enabling rapid, purpose-led innovation in anomaly detection to safeguard cross-border transactions and other use cases.

Bryssinck said, “We decided to create a centre of excellence for AI, not with the purpose to centralise all the data science activities in a single team, but to provide the tools and the methodologies and approach so we could support the data scientists spread over the various product tribes.”

The platform focuses on AI use cases that matter to customers and aims to bring responsible AI innovation at scale into Swift’s products and services. 

It leverages a common approach to data science, centralised data and AI governance, and enterprise-scalable infrastructure to transform the industry in the coming months.

Agarwal said, “The requirements that we received from Swift were quite ambitious, necessitating running thousands of AI models in parallel with high throughput and low latency. It would not have been possible to build this massive-scale application without the collaboration of Red Hat, Kove and C3 AI all coming together. In that sense, I think what we have built could become a reference architecture for the financial services community.”

Such an achievement could not have been possible without collaboration.

Artificial-intelligence

The power of collaborative development 

Swift, Red Hat, and C3 AI are working together to build a high-performing AI platform that will serve as a cornerstone for future innovations in the financial services industry. 

Kelly Switt said, “Red Hat plays the role of being the great collaborator in bringing the right resources, whether it’s from the infrastructure or from the application layer, to bring to bear for our customers.”

The partnership involves Red Hat’s technology, C3 AI’s model-driven architecture, and Kove’s scalable memory resources. 

Red Hat brings containerised infrastructure to the table, which can be hosted in various environments for maximum performance. 

Kove introduces the concept of software-defined memory, making AI-powered solutions more accessible and sustainable. 

Together, the collaboration is working to build a foundational technology platform capable of powering AI solutions for cross-border payments, amongst other financial services applications.

The vision is to develop AI at an enterprise scale, focusing on critical AI use cases and showcasing their ambitions for the future. 

The work aims to foster responsible AI and data security practices, focusing on anomaly detection to safeguard cross-border transactions.

Bryssinck said, “Anomaly detection is really a key use case for us that we are all collaborating on.”

Agarwal added, “You can think of financial services organisations as a process plant with money flowing in its pipelines. We bring over a decade of learnings in leveraging AI for predictive maintenance and reliable operations to detect anomalies and operational issues in financial services.”

Technological advancements

The new technology leverages software-defined memory, which is a cost-effective way to fully use the computing power of an infrastructure seamlessly without needing application changes.

Overton said, “[Software-defined memory is] virtualised memory…. You can take memory and put it on the other side of the building, and it will operate as if it were local memory at the same speed… What it means [from an infrastructure point of view] is you can recalibrate the servers that you have. You can have small servers that can be donated memory from a common pool in real-time.”

This innovation is set to revolutionise financial crime detection through advancements in anomaly detection and other use cases.

Overton added, “All of a sudden, a 64-gig server can become a five-terabyte server, a 10-terabyte server. You can have a graph database that’s a 100-terabyte server. There is really no limit.”

Agarwal highlighted, “To take advantage of the rapidly changing AI landscape, banks need to take a platform approach to building AI applications. Banks need a platform that abstracts away the complexity of underlying infrastructure, is scalable to meet enterprise needs, while providing explain-ability, transparency, and data lineage to meet the regulatory requirements for responsible AI.” 

While the technological innovations are central to the platform, organising them around a collaborative effort brings the best of each to the fore. 

This is exactly what Swift, Red Hat, Kove, and C3 AI are aiming to do with their platform development in the space.