DAWN OF A NEW AIR-A

SmartStream AIR has been hailed as a gamechanger – a reconciliation engine powered by artificial intelligence that has been trained on real-world data being used by global banks. We asked CEO Haytham Kaddoura, Global Head of Managed Services, Nick Smith, and CTO Andreas Burner what their (human) predictions were for the future

The beauty of artificial intelligence (AI) is that the more it’s used, the better it gets. And, after months of learning in its Vienna Innovation Lab, SmartStream’s AI has become very smart indeed; good enough to breathe new life into payment and liquidity management. SmartStream was not among the first to market with an artificially intelligent solution, and deliberately so. It has adopted a customer-first approach to developing SmartStream AIR, the software business’ latest intelligent reconciliation engine (it stands for AI In Reconciliation), which will be launched this autumn. AI is now also an integral part of TLM Aurora, SmartStream’s existing, widely deployed digital payments control solution.
“We waited on the sidelines when others were going out there to tout AI. While AI appeared to be in everything, it was diffused and ill-defined,” says Haytham Kaddoura, SmartStream’s chief executive. “People were claiming AI was being used when what they had wasn’t really AI, or their solution couldn’t use it.“SmartStream AIR is a genuinely AI-powered reconciliation engine, plus we’ve introduced AI-enabled components into our core solutions. From this year, AI becomes central to our managed services offering and
for our digital payments solutions.”

Reconciliations have long been the lifeblood for SmartStream – Deutsche Bank being just one Tier 1 client to have outsourced its entire reconciliations operation to the company –so, when its Vienna lab was set up in early 2018 to be the test bed for future products, reconciliation was identified as one of
the first AI use cases. Like many of SmartStream’s products, AIR is a plug-and-play, Cloud-based system, meaning it can be set up quickly and avoids major capital expenditure.
Kaddoura says: “There’s no question that with SmartStream AIR we have a game-changing product that achieves things no one else’s products can. It’s so much more efficient in areas such as exception management and data loading. And the major angle regarding AI is that AIR is smart enough to detect what kinds of field you are trying to reconcile and offers its own analytics to do that.“Meanwhile, we also have the benefits of AI within the Aurora family of products that manage digital payment control. This shows that, for us, AI is more than just a label; it spans everything we do and offer. It’s a feature of new products and existing ones. This is where SmartStream is evolving to.”

Arise, Aurora, for an AI age

TLM Aurora is the latest incarnation of the Transaction Lifecycle Management platform SmartStream has offered since 2006. It is designed to support the new industry standards for digital payments and covers mobile, cash and card payments, digital currencies, settlements and reconciliations. It connects with global payment system SWIFT gpi and blockchain-based networks.With payment transaction volumes exploding due to factors such as contactless technology, Aurora is designed to be scalable, cater for real-time analysis and reporting, and can support organisations of all sizes.“TLM Aurora is far more than just an upgrade,” says Kaddoura. “It’s the result of years of R&D at SmartStream and draws on valuable insights and intelligence from our clients and partners. As digital transformation increases, banks need more controls to link existing infrastructures through a
single solution.“We discussed TLM Aurora with several banks and focussed on the need for better data discovery, data modelling and data simulation. We recognised that liquidity management, reporting to the regulators and understanding future cash flows were critical on a daily basis.”

The power of the Cloud

While the use of AI in technology such as internet search engines and sat nav software is now accepted and commonplace, its use in the banking industry is far less established. Tight regulation is one reason for banks’ reticence to adopt – when customers’ money is involved, both financial and reputational stakes are high. A survey of industry players by SmartStream, together with Waters Technology, found that 26.3 per cent of them had AI live in their back-office operations, while 27.6 per cent were trialling it at proof-of concept stage. Another 19.7 per cent were considering a proof of concept, but a large number, 26.3 per cent, had no plans to use AI at all.The figures show that, for all the talk, the outlook for AI-driven systems in the financial sector remains cautious Legacy IT systems are another brake on innovation for established banks, and that is why the Cloud-based software-as-a-service model is so important to SmartStream. The firm’s global head of managed services, Nick Smith, has witnessed an increasing number of existing clients, who had SmartStream services hosted on-premise, moving to its Cloud-based solutions. This, he says, is often due to the dawning realisation that their legacy systems had reached operational limits. He adds: “Upgrading servers is an expensive proposition for a Tier 1 bank. They can avoid that capital expenditure and the additional, and ongoing expense of maintaining the hardware, by moving to the Cloud. A lot of businesses have a Cloud strategy already in place, but we can demonstrate the value we add, part of which is around cost, but it’s also about the service delivery. “There’s a real value proposition that trickles right through their organisation. And we can validate that by providing live case studies – people can talk to our clients and they will confirm exactly what we are telling them.”
Plugging in SmartStream’s AI-driven software will now see them reap the benefit of systems with far more powerful forecasting power than had been available before, according to Kaddoura.“As well as better management reporting, they’ll get better predictive analysis that can forecast how particular transactions, or cash positions, will be affected by various external and internal factors,” he says. “If a payment gets delayed, for example, how does that play out in the treasury operations of the bank? Knowing this will enable better decisions about capital deployment. The areas it touches on are multiple – operational efficiencies, enhanced risk management and capital deployment.”

It’s super Smart!

In developing its new generation of intelligent software, SmartStream has benefitted from its huge customer base of 1,500 clients, which include 70 of the world’s 100 biggest banks. It has learned by harnessing the data provided by these contracts. “We run operations on behalf of banks, and we deploy AI internally on what we run, and then realise those benefits elsewhere,” adds Kaddoura.“The software solutions we deliver are not driven by hypothetical issues we’re trying to prove, they are the result of work we have done and proved internally. We’ve put our software to the test with existing clients and it allows us to then develop our commercial products.”
Of course, data is key to the success of AI, and when a client buys into SmartStream’s AI solutions, they must also buy into its obsession with data hygiene. To process
the vast quantities of data needed to cope with modern digital banking systems, the company has developed tools to organise and store data in a standardised format to avoid a garbage-in/garbage-out scenario. At the company’s core is the SmartStream Reference Data Utility, originally created in partnership with
US banking giants Goldman Sachs, JPMorgan and Morgan Stanley. It helped the company’s first clients by providing complete, timely and accurate reference data on request. And with a history of such powerful data analytics, SmartStream was able to take the step towards AI to re-engineer traditional work models across back-office processing. Kaddoura says: “Banks have collected data ever since they existed but it’s only in the last decade that they have started realising the value in it. Having the right data in the right format, updated consistently across their client base and operations, is now the big challenge. “There have been some major reports recently that have looked at the banking infrastructure’s ability to clean and enrich this data, so that it is useful from an AI utilisation perspective. AI is going to make financial services much smarter and more efficient. It has the ability to pick up data trends much faster than any human can and that makes the banks much more responsive in addressing their own regulatory requirements, their feedback to clients and all their stakeholders.”

Something to prove

SmartStream’s launch of AIR and the AI enhancements to TLM Aurora fuel the financial sector’s potential to embrace an artificially intelligent future, which has until now been held back by low levels of internal IT expertise in AI and related fields, as well as a lack of investment following the credit crunch.SmartStream’s mutualised model provides a solution to these problems, since expertise is outsourced and the costs are effectively shared between customers. On top of this, SmartStream has a 20-year plus reputation for serving the industry to uphold.Kaddoura says: “There’s a strong level of trust that is moving us increasingly into a greater strategic role in relation to these financial institutions. I think it’s down to the changing nature of the environment that banks are increasingly wanting to work strategically with a fintech such as SmartStream.“Bank IT departments, with all due respect, are not necessarily sufficiently funded, or geared up, to address requirements from a wider industry perspective. It’s much more efficient for a fintech player, which is well embedded with financial institutions on a global basis, to do so. SmartStream is happy to be playing that role. And it means our portfolio of products and services is growing.” Smith adds: “It’s been a J-curve in acceleration because we recognise that, in this day and age, nobody wants to sign up with a vendor and then hear it’s going to take six months before they see any benefit to their organisation. Our dedicated teams do this day in, day out and, as they’ve got smarter, the automation of client onboarding has got better and faster.“It’s also important to say that we’re not an outfit making a play because we see a space in the market. There’s a real commitment here and real investment at SmartStream, to make sure our clients are going to want a long-term relationship with us.

“I spent 25 years working in Tier 1 investment banks, doing the same role as I do today at SmartStream, running the day-to-day operations,” adds Smith. “The standards we bring to the organisation and our clients are exactly what the banks experience internally themselves, including the risk and control framework and the constant drive for improvement.”

An international spread

SmartStream is a truly global company, providing solutions for financial players in every major region. So, clients are right to ask how Cloud-based solutions can meet the specific demands of regulators around the world, and Kaddoura admits that in the past this has been an ‘Achilles heel for any Cloud services provider’. However, he argues that as hosting services offered by the likes of Amazon, Microsoft and Google become more geographically widespread, those providers lay the groundwork for compliance. He says: “Some of the challenges we face are in more mature markets, but to take Amazon Web Services (AWS) as an example, its data centre in Bahrain will help with a lot of our clients’ needs in the Gulf. Other centres are being rolled out globally. “By working with Cloud services providers, we’ve helped to address these issues.”Smith adds: “AWS has a global footprint so, for instance, clients in Australia benefit from its Australian data centres and our clients in North America benefit from its data centres there.“One of the other benefits of using AWS is that we’re able to onboard clients quickly.“It can take six weeks from initiation to going live for a firm such as an asset manager with a global reach. That’s extremely fast and it’s down to AWS’s global coverage and our in-house capabilities.”

FAST OFF THE GRID

The emergence of AI in SmartStream AIR and TLM Aurora has been delivered within a mere 18 months, thanks to input from the company’s Vienna Innovation Lab.Led by chief technology officer Andreas Burner, the vision was to create a lab with the ethos of a works Formula 1 motor racing team – whereby the very brightest technologists build and test potential products away from the main business’ ‘shop floor’. Haytham Kaddoura says: “At the start of 2018 the lab wasn’t even here. But we had been watching and listening to our clients’needs closely, so we had clear ideas about what the market wanted.“We’ve been able to turn around products relatively quickly because we had concrete foundations. Unlike a company working from a concept, we already had the software solutions being used by clients, and we have developed the AI edge for them. What we can do that many rivals can’t is marry our history, our knowledge and expertise with this new technology. Our approach to building these software solutions is totally consumer focussed.” Nick Smith says that, because of this, reconciliation was an obvious area for the lab to develop an AI dimension, and it is now helping to reduce the manual, human work his branch of the business conducts.

The exceptions gap has already been closed to around one to two per cent, and the target now is to use machine learning to narrow it further. “Clients don’t really care about auto-match rates of 98 per cent or 99 per cent,” he says. “They care about the remaining one or two per cent because that’s where their work effort is spent. The AI developed in the lab sets about this, monitoring that one or two per cent and reducing those numbers. The clients then benefit, because they see fewer exceptions flowing back to their organisation.” However, Burner is keen stress that his lab is looking beyond simply automating human work. Because people with expertise in both the finance sector and data science are scarce, he argues they need to be used to their full potential. “In my lab there’s an excellent mathematician who created a machine learning model based on echo state networks – a very interesting technology to predict data,” Burner says. “It took him two months and it was amazing when we saw the results, how well it can predict around holidays and other events.” He adds that the possibilities for machine learning are continuing to grow. “You can train machine learning algorithms on thousands of levels: you can run it on your bank’s balance, you can break it down on a department level, you can break it down per currency or per country. “It will monitor and predict on those thousands of detailed levels where, as a human, it’s impossible; you would need a massive amount of people working on that.”


This article was published in The Fintech Finance Magazine: Issue #13, Page 8, 9, 11 & 12

Author: Yash Hirani

Share This Post On