SEB’s team employs AI to service alternative assets

Being the Nordics’ largest custodian, SEB Investor Services was seeing an increasing need to offer its clients services not only around securities but also the operationally awkward alternative assets. The team decided to skip the evolutionary step of manual processing and went straight for full digitalisation. Now the first participant is live – with self-learning algorithms from Google Cloud doing the heavy lifting. 

“Essentially, we took the least standardised process we could find – and decided to digitalise it,” says Fredrik Söderlund, product manager for portfolio solutions at SEB Investor Services/LCFI.

“We have served our clients for a long time around their liquid assets – the shares, bonds, derivatives, funds … – but our portfolio has been lacking a service for their illiquid instruments.” 

With the availability of new artificial intelligence (AI), SEB could now go straight for a setup that is digital at its core, rather than starting with a manual one. The common view in the industry has been that – in stark contrast to the shares, bonds and most derivatives – alternative assets are just too complex and unstandardised to lend themselves to automation. The SEB team stopped to ask itself whether, in fact, the opposite could be true? What if it is just becausethe alternative assets are cumbersome that their processes must be digitalised rather than human-staffed?

Started with a “sprint”

A process-modelling platform by software provider Appian was configured by SEB’s own staff to connect the pieces. Google Cloud supplies the self-learning data crunching tools. But with the relative ease of connecting digital services today, no traditional IT consultants were needed to staff a project. The main design considerations were drawn up in a five-day “design sprint” workshop in September 2020. Development started in January 2021, and was completed half a year later. The contrast against traditional bank IT megaprojects could hardly be more striking.

“The big projects cost big money – with large project groups, consultants and testing. But when you start from a blank sheet today, using new technology based on blocks that you assemble in lego style, with pre-defined integrations and self-testing capabilities … it gets so much easier,” says Fredrik Söderlund. 

Most of the AI-powered work is about “intelligent document processing” – starting with the extraction and interpretation of data from the broad range of emails, PDFs, spreadsheets and other files that are exchanged between stakeholders. If the system is uncertain, perhaps because a document format is new, it will loop the document to the human staff for help to match incoming data with the database blanks it should fill. 

A large mass of assets

For institutional investors, the hunt for returns and fierce competition for attractive assets are only two of the many reasons for seeking diversification and better returns outside the capital markets. The alternative assets are not new – these include forests, energy plants such as solar and wind power parks, buildings and bridges, private equity and debt, venture capital holdings etc. And they constitute a large part of the total wealth. Denmark is a Nordic leader in the space, with alternatives making up over a fifth of the institutionally held assets. But the pressure to keep down administrative costs is mounting. 

Tomas Engel is SEB Investor Services’ head of sales and relationship. 

“We see a large number of clients who wish to have these services. Many of the larger custodians are offering this in one way or another already so now we are joining them. As this is one of the more complex and manual processes for the clients, a service based on AI and a lot of knowledge is quite attractive. Also, as this field is growing substantially you actually need to be able to ease the burden on behalf of the clients to stay relevant as a service provider for the years to come,” he says. 

“And with this move, we have gone from offering nothing, straight into the future”, he adds, with a smile.

With all the billions’ worth of alternative assets under management at Nordic investment institutions, he hopes that many will be attracted to outsourcing their administration of them to SEB. Technically, alternative assets are not stored at SEB’s custody, which securities are. But if the institutional clients provide SEB with information that reflects their contracts, the services around them can still be performed. An example could be the fulfilment of payments into a venture capital fund when its general partner issues a “capital call” according to previous agreement. 

Another key value of the new service is that it can feed the aggregated information about liquid and illiquid assets into a combined overview in the client’s portfolio management system. An integration with SimCorp Dimension is already in place. 

Banking gets a new suit

So has it all been a simple and smooth process?

“Yes, it feels like it when we look back, but naturally it didn’t always feel like it during the process,” says Fredrik Söderlund. 

The story could be a text-book example of how banks across the world, like firms in many industries, are currently transforming themselves into digital-first enterprises, increasingly catching inspiration from the ways agile start-ups are working. At SEB, naturally, the new solution needed to be defended through the usual chain of committees. The growing use of cloud architectures raise new questions.

Security and data integrity are of course essential when implementing cloud-based AI solutions. Data stored and processed in Google Cloud is separated from the consumer-facing services that Google also offer. Google personnel do not have access to SEB’s or its clients’ data. Furthermore, by the earlier announced partnership between SEB and Google Cloud, they also have a solid security fundament together. This is a good example of how this partnership delivers new, innovative and secure services.

Based on the Appian platform, the SEB team is able to configure business rules and workflows and data structures rather intuitively, without needing to hire expensive coding experts. This type of “low-code” approach is rapidly becoming popular across the financial services industry. 

Keep it small – and do it faster

Fredrik Söderlund and Tomas Engel believe that many institutions, possibly also among its institutional clients, could benefit from a similar search for new processing possibilities. Just how solutions should look must depend on the situation at each company, but in SEB’s case it has worked well to move towards “task-based” principles, the low-code approach, a small cross-functional development team and the use of artificial intelligence (calibrated by large data from internal asset administration in other parts of the group).  

“To do something like this in a financial institution, you do need to radically reconsider how you work,” says Fredrik Söderlund. “And sometimes, it can be better to be fewer people.”

Tomas Engel agrees: 

“Today, sometimes all it takes is a small group that thinks new and smart.”