R.Babakin - Fotolia

Danske Bank and IBM bring AI to banking IT

Artificial intelligence partnership is designed to reduce the risk of major IT outages and keep the bank’s customers happy

IT outages can cause major headaches for banks. As well as damaging customer goodwill – something few banks can afford to do – such outages are increasingly being scrutinised by regulators and governments.

Patience in some sectors is wearing thin, and sanctions, including financial penalties, may become more commonplace. The fallout from the TSB outage in the UK in 2018 is still reverberating around the banking industry, and it is far from being the only recent serious event.

The more often such outages occur, the greater the risk of banks losing that most precious of human commodities – trust. As long as bank depositors can access “their” money, trust, both in the financial system as a whole and in a given bank in particular, is likely to remain intact. When they can’t, or fear that they won’t be able to, as happened with Northern Rock in the UK at the start of the financial crash of 2007/2008, a bank run can swiftly result.

That is even more of a risk today, when news of banking problems can spread much more quickly, thanks to social media and 24-hour rolling TV output.

It is against this backdrop that Danske Bank has partnered with IBM to use the latter’s Predictive Insights technology to try to predict potential outages before they happen, so they can be minimised or prevented.

Craig Ian Alexander, senior vice-president and co-head of IT operations at Danske Bank, told Computer Weekly: “Banking is complex and AI [artificial intelligence] is something that helps a human being to deal with complexity in a more efficient manner.”

According to Alexander, the partnership with IBM is the latest phase in Danske Bank’s Cognitive project, which began in 2017. “The initial phases were primarily focused on scoping and planning, as well as customising the solution for our specific needs,” he said.

It took about four months to implement the technical side and select the specific data to be used for prediction, but the more time-consuming phase is training the system to know what to look for and then act accordingly, he said. Based on IBM’s own experience, this can take many months to achieve, said Alexander, adding: “This lengthy process should deliver real service value.”

But such value is not always easy to measure. Danske Bank is using two key metrics – customer impact availability (CIA) and median time to restore (MTTR) – and hopes both metrics will be improved by IBM’s Predictive Insights system.

Read more about digital banking in the Nordics

Alexander said it is in MTTR where Predictive Insights can have the most impact. “The earlier we can detect a potential incident, the better able we are to respond and therefore resolve it,” he said. “Predictive Insights is aimed at giving us the ability to fix an issue prior to an incident occurring, but in certain circumstances this will not always be possible.”

Minor outages, affecting small numbers of customers for a relatively short period of time, can be annoying but are rarely seriously damaging to a bank’s reputation. Within certain boundaries, customers understand from their own experience of technology that IT doesn’t work perfectly all the time.

However, longer, more widespread and severe outages really have the potential to dent a bank’s reputation and even affect its long-term outlook. “Our CIA has been at its highest ever for the past two years and this will be maintained more easily with a further reduction in our number of major incidents,” said Alexander.

But none of this has yet been proven in reality. “These are just the initial attempts to apply the technology outside the IBM environment, as the technology is still very new,” he said.

There will inevitably be teething troubles, but those should be ironed out before the Predictive Insights system is given a more central role in identifying and responding to potential outages, he added.

One of the main issues IBM has had to deal with is Danske Bank’s highly dynamic server environment. “We have also raised a requirement to build the service element into the anomalies that we are monitoring, so that we could clearly understand the importance of them from the customer perspective,” said Alexander.

High-quality implementation

This was different from previous IBM implementations and has required additional development work, he said. “These elements have taken time, but we see them as crucial for a high-quality implementation.”

Alexander is optimistic that the system will make a big difference to the bank. “Applying AI helps us to be faster in serving customers, solving their issues and focusing our resources on creating new solutions for the customer,” he said.

Asked whether such a system might have helped prevent last year’s TSB problems or other recent outages, Alexander said: “We see great potential in applying AI, but we do not have enough insight into that specific case to say with a high degree of certainty whether it could have made a difference.”

Alongside the Predictive Insights implementation, Danske Bank is also making use of IBM’s Watson in its IT service desk, with an AI-powered chatbot to help resolve IT issues for internal bank staff. Both tools are key factors in making IT support faster, more automated and more efficient, said Alexander.

“During the upcoming year, we want to spend time training these solutions as much as possible to derive the maximum value,” he said. “The next step could be scaling these solutions to other environments and other support areas.”

Danske Bank’s venture into AI might sound at odds with the traditionally conservative image of banking, but it’s a sign of the times. Banking is now an IT-driven business, and banks that fall behind on technology may find themselves losing their competitive advantage.

Looked at in that light, Danske Bank’s partnership with IBM is a logical step, even if the true value of the new system in terms of customer satisfaction – and trust – has yet to be demonstrated.

Read more on IT for financial services

Data Center
Data Management