Industry
Banking
Banking Services Transformation: Unlocking Exponential Value
Solution
Consulting
TCS Consulting Services to Achieve Success in the Digital Era
Cognitive Business Operations
Unlock Business Agility and Future-ready Operations with TCS
Highlights
- As CFOs explore new technologies to reimagine the finance function, they are keeping track of how generative AI (GenAI) evolves.
- GenAI has the potential to help CFOs better manage profitability, margins, and costs.
- Financial institutions must adopt a use case-based approach to GenAI and run proofs of concept (PoCs) to create workable prototypes that deliver business value.
On this page
Abstract
The financial services industry has always been at the forefront of embracing new technological innovations to keep pace with the changing business ecosystem.
Generative AI (GenAI) is no exception to this trend. The advent of GenAI offers valuable opportunities to enhance productivity in financial services. That said, there are differing views on both sides of the spectrum, spanning acceptance as well as opposition to adoption. We believe that a cautious approach is the right way forward. We highlight areas where GenAI can be leveraged in the finance space to improve key functions of the chief financial officer’s (CFO) office.
Art of the possible
GenAI comes with superior learning models and the ability to assimilate complex content and produce insights that mimic human intelligence and knowledge.
It opens up a plethora of opportunities to resolve complex problems and conduct repetitive tasks with an enhanced degree of precision. Its capabilities go beyond what is offered by existing robotic process automation (RPA) and machine learning (ML) technologies. GenAI models can be continuously trained to learn and relearn, which enhances their learning capacity multi-fold. In contrast, RPA and ML have limited learn, auto-learn, and relearn capabilities as the benefits realized through them tend to plateau over a period of time. Compared with other data crunching tools, GenAI has a higher capacity to consume huge amounts of data and deliver value-generating insights and greatly improve the quality of decision-making in the financial services industry.
How GenAI helps CFOs
Several CFO-centric business functions can benefit immensely from GenAI.
Given that the finance function is laser focused on profitability, margins, and costs, GenAI has the potential to complement CFOs in achieving these priorities.
Balance sheet management and profitability: CFOs of financial institutions must factor in multiple conflicting parameters and criteria while finalizing the balance sheet. GenAI enables finance teams to consume large swathes of market data that is used in balance sheet planning and extract insights and synthesize data for better decision-making. With pretrained parameters, GenAI can learn and generate insights at speeds and precision that existing technologies cannot match. Consequently, CFOs can get timely access to superior insights without spending time on extracting them, thereby empowering them to make smart decisions.
Profit and loss and balance sheet substantiation: Analyzing trade level transactions and performing variance analysis on different parameters such as price fluctuation, deal level profitability, intra-day profit or loss, and end-of-day valuations require review of multiple documents covering deal data, trade level transactions, Greeks, and so on. Currently, this process involves manual intervention, compromising accuracy. GenAI can read through troves of background content and leverage large language models (LLMs) as well as pre-trained and proprietary models to synthesize highly accurate commentary, eliminating human intervention so that business teams only need to scrutinize the final commentary.
Financial risk management: The CFO function of financial institutions monitors exposure and deposit concentration and oversees compliance with International Financial Reporting Standards (IFRS) and Basel regulations. GenAI tools can scan market reports and perform sentiment analysis on news feeds and read through regulatory text and summarize findings. More importantly, the tools can suggest alternatives to mitigate concentration risk in the existing credit portfolio. This will help flag events that need immediate attention and alert business teams to initiate action.
The US Fed’s stress tests do not factor in select scenarios such as an increasing interest rate regime, percentage of uninsured deposits over the Federal Deposit Insurance Corporation (FDIC) limit of US$ 250,000, and banks holding a larger amount of treasury bills in the held-to-maturity (HTM) bucket. As a result, aspects such as deposit concentration and maturity mismatch are not factored into stress tests. In our view, this combined with regulators’ failure to foresee the potential of such excluded scenarios to trigger a crisis is one of the reasons for recent bank failures in the US. Existing technologies lack the capability to extract the requisite data from banks’ systems to run such new scenarios. Deploying GenAI can help simulate such scenarios and potentially avert a future crisis.
Treasury operations: Treasury division functions such as investment planning, funds transfer pricing (FTP), and nostro-vostro reconciliation involve research of different sources of information. Existing processes are largely manual, and significant time and effort are expended to perform treasury operations. Consequently, time available to analyze, review, and take the right decision is limited. Armed with pre-trained models and the ability to assimilate large data sets, GenAI delivers accurate, predictive insights to treasurers for decision-making, eliminating time spends on manually reading reports and drawing insights.
SOX compliance: Corporate failures and fraud resulting in massive financial losses led to the introduction of the Sarbanes-Oxley (SOX) Act in 2002. The objective was to improve risk management and governance. An independent audit board, the Public Company Accounting Oversight Board (PCAOB), was formed to scrutinize the audits conducted by banks and prescribe corrective action. However, compliance is complex and still poses challenges. Internal financial controls mandated by SOX involve review of troves of financial documents, process manuals, standard operating procedures, checklists, and so on to assess financial controls. Traditional tools and technologies have many pitfalls, owing to which SOX, internal financial controls, and audit processes in banks are imperfect. With its ability to assimilate large swathes of data, GenAI can help the CFO function to overcome challenges in traditional methods, deliver superior outcomes, ensure efficient monitoring, and achieve effective regulatory compliance.
The way forward
GenAI is memory intensive and requires heavy compute power.
This means existing pretrained models may not be suitable. Financial institutions will have to make significant investments in compute infrastructure and training data repositories to deploy Gen AI technology. We believe that banks must tread cautiously in identifying training data sets and finetuning the existing commercially available models that would be a part of the GenAI ecosystem.
Before embarking on implementation, CFOs will need to resolve some key questions.
- Do the spends on GenAI justify the returns?
- Can the costs be capitalized and amortized over the years?
- What is the shelf life of GenAI models?
- Is GenAI climate-neutral? What are the environmental, social, and governance (ESG) implications of GenAI adoption?
Another point to be noted is that finetuning existing models comes with the risk of harmful training outcomes and the potential for inadvertent biases in model training. The solution lies in creating proprietary data sets to train models, adopting a cautious approach to finetuning pre-trained models, building industry consensus on how to use GenAI, and nudging industry bodies to ensure ethical use. In addition, training must be tailored to specific functional domain(s) encompassing trillions of parameters, diverse asset classes, and data taxonomies. Human oversight is key to ensuring that errors and omissions with the potential for biased decisions are eliminated—this responsibility cannot be delegated to a machine. A human being in the CFO function needs to be held accountable for the consequences of inadvertent errors and omissions.
GenAI is a classic example of technology outpacing regulations—there are as yet no regulations to govern GenAI adoption and use in the financial services industry. Given the ambiguity, financial institutions must self-regulate, building consortiums to define guidelines. They must voluntarily comply with the guidelines till regulations eventually catch up with industry readiness to explore and embrace GenAI. Bank CFOs must keep in mind that they cannot drive the conversation on reaping the benefits of GenAI on their own. As with every big initiative, industry and rule-making bodies must collaborate, which needs a concerted effort from lawmakers, elected representatives, industry bodies, and banks.
In light of the above, we recommend that financial institutions adopt a use case-based approach and run proofs of concept (PoCs) to check if ideas can be transformed into workable prototypes that deliver business value. The learnings from the PoCs can be leveraged to determine if GenAI technology should be pursued for other use cases. Having said that, use cases will continue to evolve, and arriving at a consensus will require multiple iterations. Further, regulations on data localization, privacy, and hosting as well as geography-specific privacy laws must be borne in mind while building utilities or backend software using GenAI technologies. CFOs must consider inputs from the CIO and CTO functions on the most relevant and suitable use cases. They must then formulate a cross-functional task force to embrace GenAI and deal with the change management ramifications. The GenAI utility must not be given the autonomy to make decisions—human judgment must inform the final decision.
In summary
GenAI is not a silver bullet that will resolve all business challenges that CFOs grapple with.
However, some financial institutions have embraced GenAI in a limited way for their day-to-day internal activities, running pilots for specific use cases to determine if large scale adoption is feasible. As CFOs explore new technologies to reimagine the finance function, they must keep track of how Gen AI evolves. Once adoption picks up pace, quick action will be needed to forge ahead of the competition.
