The notion of straight-through processing (STP) — of automating every step and eliminating manual work – has been a guidepost to driving efficiency, minimizing errors, and optimizing operations since the early 1990’s. In financial services, it has transformed many transaction processes including Securities processing where STP rates for Settlements have been 90+% for many years.
However, as STP rates have gotten higher, many transaction processes have hit a point of diminishing return for further STP investments and rates have stayed stubbornly stuck in the 90+% range for years. More insidiously, STP transaction processes have encountered exponentially higher costs of change as the complexities of last mile automation collide with the increased frequency of business change. Banks especially are recognizing that STP is no longer the promising strategy it once was for quick ROI at the same moment that the digital age is driving an explosion of transaction volumes.
What once seemed like a few missing points of STP translates to multiples of transactions a day that demand manual intervention and create operational risk. Meanwhile, the same digitization brings compressing margins and the pressure to substantially reduce manual touch points.
So, what’s to be done? Many are looking to the predictive and generative power of AI to help continue the STP journey. And many are eking out higher STP rates with this approach, but in doing so are encountering an even higher cost of change (as AI algorithms increase complexity) and considerable resistance from compliance, risk and audit regimens that set high thresholds for observability, explainability, and traceability.
Enter AI Operationalized (AIO)! We find that AI integrated well into human-in-the-middle workflows can dramatically improve productivity and lower operational risk to enable STP-like rapid returns while keeping current regulatory and risk control frameworks largely intact. While AIO immediate objective is not STP, it offers orders of magnitude improvement in productivity and offers lower risk at much lower costs and higher speed of implementation.
Achieving rapid returns through AIO though does demand a fundamentally different approach to scoping automation efforts than traditional STP. The difference in approach is illustrated in the graphic below:
While most STP automation efforts focus on eliminating manual touchpoints altogether, AIO focuses on solving for manual touchpoints that are labor intensive, especially those that demand human judgement and intense focus to make complex decisions and sift through mounds of unstructured data. As such, when we initiate AIO work, we start by triaging manual touchpoints to find the most labor intensive processes and examining them for opportunities to use our Predictive Exception Processing (PEP) and Unstructured Data Processing (UDP) AIaaS platforms to create targeted solutions.
Then we focus on how best to combine human intelligence with artificial intelligence (HI + AI) to preserve the highest levels of quality and controls in the transaction processes while grossly reducing manual efforts and commensurate risks for rapid ROI.
We believe AIO is a necessary approach, picking up where STP left off to help financial services operations move AI from a promising technology to a highly valued capability able to transform professional lives and drive massive ROI at speed.