Similar arguments apply to machine learning, which can be a boon if there are sufficient volumes of input data from which the machine can ‘learn’, such as data generated by payments and transactions. In a Jersey context, this learning could, for example, determine whether certain structures or types of information exist in a document. Or it could be used for the huge volumes of data analysis in M&A activity or private equity investment. However, the business case for machine learning solutions where the scales aren’t evident is generally niche.

The challenge for Jersey-based financial services firms is to understand where AI can benefit them and whether scale and data availability can be achieved when operating across multiple jurisdictions.

It may be that different regions, business segments and cultures make this impractical – or they introduce the risk of spurious results. On the other hand, a firm could find that the scaled benefits make AI adoption more practical than its sole use in Jersey.

Another consideration is the fact that many of Jersey’s clients have some form of financial relationship with other jurisdictions – and it may be that the other jurisdictions are better placed to leverage economies of scale in the use of AI. This could create an imbalanced client expectation of turnaround times or fees, making this another vital factor in deciding whether AI solutions should be introduced locally.

In our introduction to AI, we discussed why now’s the right time for AI to thrive. Globalisation is a key force driving change in the financial services industry, with several related consequences, each of which can trigger the rise or fall of industry segments as international market forces or regulatory change open up sectors or close them down.

In particular, globalisation is driving the consolidation of players in the industry as backers aim to build bigger, better businesses to take advantage of scaling opportunities.

This consolidation of businesses results in the redesign of operating models, giving rise to a trend towards outsourcing functions to lower-cost jurisdictions or centres of excellence.

Influenced by globalisation and changes in international standards, regulation, especially on the ground in Jersey, also materially impacts what businesses need to do and when they need to do it by. However, regulation has sometimes caused a reversal of what once seemed an inevitable outcome. An example of this is the moral debate around privacy and AI such as facial recognition or using personal data for marketing. While the use of these technologies had long been anticipated, in certain cases it has gone so far as to feel intrusive and uncomfortable, and adverse consumer reactions have limited or even reversed their adoption, driving positive change.

These factors play a significant role in shaping proactive and reactive business strategies and are rightly placed at board level. What’s often underexplored is the role AI can play in helping firms meet their strategic objectives.

Let’s turn to the technologies that are impacting the financial services industry in Jersey. Top of the list are the mainstream back-office systems used to manage the administration of processes across each of the main industry segments.

Capgemini estimate that 90% of European and North American technology budgets are spent on managing and maintaining legacy systems. Implementing or upgrading to later/better versions, or even consolidating costly technologies, has the potential to improve the efficiency or effectiveness of managing an organisation.

Also significant are the infrastructural technologies that underpin or complement  back-office systems. Among these are the so-called ‘cloud’ providers of outsourced computer servers, including all of the related productivity tools such as Microsoft Office 365, which means that no business needs to have its own internal centralised servers.

When you consider the sophistication of the internet, including interfacing technologies such as web services standards and the emerging internet-of-things, along with global low-cost technology suppliers, the direction of travel is very clear.

With a combination of consolidated systems, upgraded technology, cloud-based solutions and better interfaces between systems, you have strong foundations for the implementation of AI solutions.

However, VentureBeat AI predicted that just 13% of AI projects will make it into production, citing a lack of data strategy, overly complex projects, the inability to deploy models due to resources, and a hesitancy to invest from senior management.

Compliance and fraud detection

Solutions can offer incredibly high accuracy and ‘always-on’ availability, which is useful given the 24/7 nature of global financial transactions. Tools can be used to collect and analyse thousands of transactions in mere seconds against a specific set of pre-defined rules. Where a transaction fails to meet the rules, and therefore may be deemed ‘suspicious’, a report to the compliance team can be generated instantly, at which point human intervention can handle the issue. AI can supplement identifying patterns in vast amounts of data and recognising potentially fraudulent transactions more quickly and accurately than human staff.

Data entry, capture across multiple systems

During the client onboarding process, many firms require a new client’s data to be captured and entered into multiple systems. A robot can automatically transfer information from one system to another – for example, from a back-office onboarding platform into a customer portal platform. AI can then enable the decision making process inline with business risk appetite and initiate the downstream data workflows and processes.

Reconciliations and validations

Most organisations will spend significant amounts of time manually checking and matching transactions and preparing and posting journals, making this an ideal area for Intelligent Automation.

Firms such as Duco and Xceptor provide Machine Learning powered reconciliation solutions, allowing data from disparate sources to be automatically ingested and compared, with installations providing an ROI within weeks.

Intelligent Automation can be used to complete tasks such as clearing trades, regulatory reporting, carrying out order research and resolving discrepancies. In some cases, functionality should be able to automate entire end-to-end reconciliation processes without the need for any manual intervention, but significant efficiencies can still be made where automation is around 90%, enabling staff to dedicate more time to complex cases and investigations.

Phishing and social engineering

Software that can understand language, tone and user behaviour to prompt when an activity or recipient is outside the norm and the user may be about to send information to the wrong person.

Data Loss and GDPR

With data the lifeblood of AI, firms increasingly hold larger amounts of personally identifying and sensitive information. This leads to an increased risk, and often this can come in the form of poor data protection. AI can be used to dynamically classify client identifying data, and prevent it from being sent or transmitted in an insecure fashion.

Network analytics and threat detection

Technology implemented that uses machine learning to understand what normal usage looks like across your technology stack and from this identifies suspicious changes in behaviour and activity, identifying and locking potentially compromised staff accounts and driving proactive alerts for investigation. Many vendors and service providers, especially those with cloud-based technologies, pool threats seen across several of their clients to identify patterns and act accordingly with updates and fixes

Many more examples exist; as with any cybersecurity strategy, the priority for firms is to understand what needs protecting and find the best technologies to provide this protection.

Machine readable AML handbooks

Virtually all regulatory jurisdictions make their regulations available online, although formats can vary from PDFs to HTML pages and text feeds. Rulebook ingestion can help financial services firms keep up with changes in regulations and refresh their processes based on these changes.  Learning technology can be taught to understand the meaning behind the regulations and categorise them for ingestion by the governance, risk and compliance platforms that allow firms to evidence compliance with these regulations.

For firms that operate across multiple jurisdictions, there’s an additional benefit with technologies able to identify scenarios and the regulatory stances of different operating locations as an enhanced way of ensuring governance, risk and compliance procedures are acceptable across the entire business.

Credit and decisions

Credit departments review thousands, even hundreds of thousands, of annual reports every year to assess the credit quality of clients and potential clients. The process by which a bank transfers information from a borrower’s financial statements into the bank’s financial analysis program is a mainly manual process, employing armies of staff with the knowledge required to read and extract key data from the annual reports.

Onboarding Example

Onboarding is an important challenge across the financial services industry, with speed of execution and user experience being key success factors. The inclusion of AI to undertake liveliness checks (which can be completed remotely) can help ensure the entity being onboarded is who they say they are and confirms their link to complex structures, which is currently a critical bottleneck.

Reconciliation Example

Reconciliation within the back office is not restricted to books and records platforms but increasingly validating multiple sources, inputs and outputs from processes across the client lifecycle. The volume of data both manually checked and subsequently exception managed is material and growing across middle and back office controls.

However, in the majority of cases, measuring success will typically be achieved by measuring one or more of the following:

Revenue or cost – increased profitability or cost savings / number of full-time equivalent (FTE) employees to give the figure per FTE

Payback period – typically linked to cost per FTE figures, the period of time by which the total investment has been repaid based on the realised benefits

Return on investment – realised benefits / total investment expressed in percentage terms

A back office example here would be addressing Error/Right First Time Rates’ – The average number of mistake or issues made during a manual process, for example re-typing mistakes during onboarding.

So, whether the AI technology used in financial services is innovative or not, it can be planned and measured using known methodologies, and quantified alongside other strategic priorities.