AI and machine learning spearheading rapid change in data reconciliations
With the demand for reconciliations spanning ever higher volumes and complex areas and with the increasing application of artificial intelligence (AI) and machine learning (ML), many firms are taking big strides in improving operational efficiency and decision-making. We spoke to SmartStream’s TLM Reconciliations Product Manager, Robin Hasson, about the trends and opportunities.
Change in the broad financial services reconciliation space is coming thick and fast. As a long-term specialist and market leader, SmartStream is well placed to understand the shifting requirements.
On the one hand, says Hasson, there is considerable demand for very high volume reconciliations, particularly in payments processing. A lot of the tech-savvy new entrants have fast and efficient reconciliations as part of their core business model, with an emphasis on straight-through processing and low costs.
At the other end of the complexity scale, Hasson highlights an area such as derivatives contracts, where there might be hundreds of data elements. Here, the emphasis is likely to be on data quality. Being able to evaluate data quality across different sources to highlight low quality or missing data points is vital for operational accuracy and reporting, and it also creates opportunities.
Having run-time access to analyse and pivot data quality provides insights that can lead to business re-engineering, both internally and when choosing your external partners. Trends of data quality are visualised by product, market and broker via multiple presentation options, including heat maps, to provide the valuable business insights.
As well as traditional areas of reconciliations such as cash, nostros, internal finance, funds and portfolios, and derivatives, Hasson feels that more and more there is a need to bring “spreadsheet departmental reconciliations” into the fold.
The overarching aim is to remove key man dependency by using a single platform and empowering the business to be more independent, a need that has only been made clearer during the last year or so due to Covid, with the disruption this has brought to workforces and processes.
The dual challenges of increased volumes and complexity mean reconciliation tools have needed to evolve, with an important part of this being ever greater harnessing of AI and machine learning. “We have invested significantly as a firm in dedicated experts,” says Hasson. AI/ML is not a panacea, he points out. It is important to understand the strengths of the technology and where it is most suited, but where this is the case, there is a lot of potential.
SmartStream’s Innovation Lab has been conducting pilots to fast-track and prove high-value AI business cases. Hasson adds: “We have been conducting Proof of Concepts with some of our largest customers, using their data with our AI components and working with the data scientist team to evaluate the cost and efficiency benefits.” Hasson is quick to emphasise that these are not heavy, high-cost engagements, so that users gain fast results.
SmartStream’s recently unveiled Affinity, its latest AI module, is embedded in the company’s reconciliation offerings and basically observes the users’ actions and establishes its own understanding of how records match. Affinity acts as a virtual user to support businesses dealing with large amounts of manually matched data – the more it observes, the more accurate it becomes, boosting matching rates.
Users are provided with information about the matching, including the AI confidence rate and graphical explanations of how the predictions are made. “This gives the business the confidence they need to rely on Affinity – no one trusts AI immediately – and they can then decide how quickly they want to convert AI predictions to fully automated matching,” says Hasson.
For some areas, says Hasson, there can be exceptional benefits for complex data sets, equating to major savings and operational efficiency improvements when it is across thousands of matches per day for even a single line of business. Proof of Concept studies have had estimates of at least 20% cost savings for users’ reconciliation business.
Other AI modules cover aspects of onboarding and data enrichment. Using AI to analyse datafeeds and recommend the mapping of data provides efficiency gains when creating new reconciliations types. And there is potential to improve both MIS and match rates using the data enrichment capabilities of the software, where aliases and look-ups to identify brokers, for example, can be learned from existing data. The correction of data at the point of load delivers operational benefit while also breaking away from the previous need to set up static rules.
Reconciliations is also certainly not immune from the overall move to legacy transformation, which typically includes some aspect of cloud hosting. The shift could be classed as Reconciliations-as-a-Service. “We are seeing a number of clients adopting a SaaS model, whereby they hand over maintenance and management of the software, with the potential to expand operational tasks as a fully managed service,” says Hasson. The shift can bring massive cost savings, he says, as well as benefits such as ensuring the customer is always on the latest version of the system.
SmartStream has 40 years of experience in this sector and arguably the current speed of change is as rapid as anything that has come before. The potential is there for large leaps forward in multiple areas, across both traditional areas for reconciliations as well as new ones.