Sanjay Prasad, Head of Capital Markets and Insurance, TCS BaNCS – North America

Straight-Through Processing has been the holy grail for both Cash and Securities Operations for the past three decades and still eludes us to-date. Combined with AI capabilities, it has the potential to elevate Securities Operations to its next paradigm. How can the industry harness the power of AI more efficiently? Let us look at a few cases and causes.

Straight Through Processing (aka STP) goes back as far as one can look. SWIFT was born in 1973 out of this very desire to standardize, simplify and speed up global financial services transactions. Fast forward to the early ‘90s, where we had GSTPA which was the next boldest and purest attempt in that direction. Like every great idea, GSTPA met a pre-mature exit largely because, standardizing means picking one’s over the other’s and we all know how easy that could be. All these, coupled with other systematic challenges, would perhaps explain why STP still, and perhaps will continue to, elude us. AI presents a new paradigm, a game changer of sorts to further the journey towards a leaner, meaner set of operations for the Financial Services Industry. What will it take to harness the power of AI to realize the vision? The right approach to solving this problem is perhaps the most critical aspect to consider

Product managers for both Securities and Cash Processing systems - be it post trade processing, clearance and settlement systems, asset servicing systems like corporate actions, or cash processing systems like DDA and Payments – are always on the lookout for opportunities to improve their technology infrastructure and make it more risk aware. As the use cases fall into the hands of a product manager, their very first instinct would be to deconstruct the transaction lifecycle, somewhat like that of a car owner wanting to customize their fine-tuned AMG engine with a power boost. More and more attributes of the transaction are being parameterized, which are then used to configure (read fine tune) the system behavior to the most optimal performance, while data is passed on to the cache, in-memory processes and so on. A combination of these parameters is further permuted to bring risk awareness. This approach has been prevalent and proven effective for a long time, and many of the technology providers and captive technologists are continuing with such an ‘if-then-else’ approach for their systems to make them more efficient and resilient. Needless to say, the benefits of such an approach is slowly and progressively becoming incremental, at best, and nowhere significant enough to create that ‘x-factor’. Furthermore, the above parameterization is adding an unintended burden on to the software system, making it less agile, hard to implement, and is also affecting its performance.

This warrants the question ‘Should Transaction Processing solutions be bundled with Intelligence?’. Enter RPA, which is often confused and excused for AI adoption. RPA is as much AI as a drive on the Daytona 500 track is that on the Nϋrburgring. RPA is mostly just simple automation based on a set of predefined conditions.

So, is there a different approach of infusing intelligence to the back-office systems?

An interesting use-case concept came up in one of the brainstorming sessions regarding Asset Servicing solutions, where it was asked if the software could project the cashflow for corporate events. Even before the question finished, in came the answer, “YES!!, we can project cashflow as soon as the event is announced and keep projecting it incrementally as the holdings in the portfolio change”. Subsequently, they tweaked the question asking ‘if the system could project the cashflow for events which were not announced yet, given there are enough historical data points to predict that event?’. Without belaboring on the detailed strategy on how it could be achieved, it would be sufficient to say that the product specialists felt vindicated explaining how, by generating dummy transactions, the team could solve the problem. This interaction highlights one of the most evident roadblocks towards the infusion of intelligence at the Back-office – ‘The Straight-Thru mindset and approach to solve a quintessential Intelligent/AI use case’. Not only could AI tools address these predictive use-cases in a blink, but it will also allow the core Transaction Processing systems to remain light-weight and efficient. On the other hand, the Straight-Thru approach to solve the above ask via and within the limitations of an STP system will create new problems making the system code-heavy, complex and difficult to implement and maintain. In short, the complications will probably outweigh the benefits.

From simple predictions like that of a settlement failure to complex ones viz., the likelihood of another Yahoo-Verizon merger, the throughput and staffing issues thereof are all examples of issues that can be best, and in some cases, only be resolved by The Intelligent (AI) mindset as opposed to The Efficient (STP) mindset.

The above is perhaps just one roadblock and being tied to human behavior, it may be paradoxically very easy or alternatively, very difficult to solve. In these times, where diversity is being revisited in many ways, perhaps a different kind of diversity may have the solution to this issue, which would be including design thinkers and AI geeks within product development teams along with technical designers, business analysts and domain SMEs. This will hopefully ensure that the products are leveraging the power of The Straight Through and The Intelligent in the most optimal manner.

Disclaimer: Views or opinions represented in this blog are based on author’s own research and does not represent TCS BaNCS