AI in the back office: time for action
September 13, 2022

Banks’ back offices are inundated by data. Advanced technologies such as AI and machine learning could help firms process and understand this information more efficient.          Yet the industry remains cautious about adoption.

A vast river of data flows in and out of banks’ back offices. Yet many firms, hampered by ageing IT systems and inefficient manual processes, neither completely understand nor exercise sufficient control over this deluge of information. As a result, firms are plagued by avoidable operational losses and also fail to benefit fully from their data. Piling added pressure on banks’ operations is the explosion in digital payments, given extra impetus by the pandemic. Increasing processing speeds is a must, but many firms struggle to meet the demand for ‘instant’. 

Skills dependency represents another headache. Important operational knowledge is often lost when staff leave, and tools are required to prevent the haemorrhaging of corporate intelligence. Freeing up staff from repetitive, low value activities, in order to carry out higher value tasks, is also essential. Greater agility, efficiency, accuracy and cost-effectiveness are necessary if banks intend to achieve superior margin, increase business volumes and retain customers. At StreamStream, we believe that innovative AI and machine learning technologies have a pivotal role to play in helping firms meet these goals. AI and machine learning are ideally suited to handling complex data sets. They can detect subtle patterns in huge volumes of data, in a way the human eye never could, facilitating  understanding. Their ability to tackle vast quantities of information also means they can inject much-needed speed into back office processes. They offer banks the opportunity to unlock the true value of their data, turning what is currently a burden into a valuable asset.

SmartStream has created sophisticated, AI-enabled solutions which transform the way firms manage and reconcile data. Our systems combine multi-year operational experience with AI to deliver actionable information. They boost automation, helping banks to improve service levels, stem operational losses, achieve superior margin management, and perform more accurate data analysis. These products are already in use, and offer proven cost-efficiency benefits. 

Three years ago, SmartStream founded its Innovations Laboratory. The purpose of the laboratory is to apply AI and ML to the specific business issues faced by our clients. It  collaborates with customers – including Tier 1 banks – on proof-of-concept projects, to identify high value business cases where AI can create proven cost and efficiency benefits. By using our AI technology in combination with banks’ own data and processes, these institutions are able to see for themselves the incredibly positive results AI can bring.

SmartStream Air is another important development for us. An AI-based application, Air  transforms reconciliations onboarding and processing, reducing tasks which traditionally take weeks or months to a matter of seconds. SmartStream Air requires no training or configuration – users simply upload raw data to the application, in pretty well any structured format. SmartStream Air then matches the information, using unsupervised AI. Transactions can be streamed in real time, a useful asset when processing digital payments. The application handles complex data sets rapidly and accurately, also lowering dependency on specialist skills, such as configurational ones.  

We have also developed an ‘observational learning’ component, ‘Affinity’, that learns from the manual matching behaviours of human users. As well as forming part of SmartStream Air, it is being embedded across our existing solution suite, including in the technology used by our managed services arm. A recent project, carried out in collaboration with a Tier 1 bank, to integrate Affinity with the bank’s TLM Reconciliations Premium platform, demonstrates Affinity’s ability to boost efficiency. The bank had high levels of automation but wanted to tackle a residual pool of complex manual matching. Affinity was used to deal with this final percentage of manual matching, creating potential savings. For large firms like this, handling huge transaction volumes, such an efficiency gain can represent a cost reduction of millions of dollars. 

Importantly, by learning from the matching activities of skilled operational staff, Affinity preserves vital corporate intelligence, preventing it from being lost or disrupted by events such as staff absence or holidays. Its ability to learn rapidly from human users also frees up staff to concentrate on higher value tasks.In conclusion, AI and machine learning offer banks a potentially transformative way of overhauling their operations. Yet the pace of adoption remains slow. Given the many and relentless pressures they face, banks can surely no longer afford to hesitate. With reliable, powerful solutions now available, the financial industry must overcome its reservations and embrace these new technologies, taking advantage of the benefits they create.