The advent of Generative AI (GenAI) heralds a transformative paradigm shift in the dynamic realm of banking. Initially met with skepticism and apprehension due to concerns about potential effects on reputational risk and regulatory compliance, the banking sector was guarded in embracing GenAI technologies. Certain institutions took precautionary measures, going to the extent of restricting access to platforms like ChatGPT.
Over time, however, the industry has recognized GenAI’s potential to revolutionize operations. Adopting a cautious and gradual approach, banks have begun integrating GenAI, focusing on three main areas of increasing complexity (and progressive regulatory scrutiny): enhancing employee productivity, improving complex operations, and developing consumer-facing applications. This journey has involved both the creation of new GenAI-driven solutions and the adaptation of existing technologies to meet the distinctive requirements of the banking sector.
While the integration of GenAI in banking has shown promising advancements, addressing the concerns and misconceptions, particularly regarding security and data accuracy, is essential.
Concerns and misconceptions
By now, we are well versed in the challenges of GenAI. However, this has resulted in several misconceptions centered on security and accuracy. Many of these challenges already have solutions:
|Indirectly identifiable information
|Use data masking techniques to hide accents in transcripts that hint at any leading identifiers that could lead to bias, e.g., ethnicity or regional details inferred from mentions of local stores or partial addresses in transcripts.
|Securing proprietary information
|Various encryption methods protect internal strategies, methodologies, or intellectual property and prevent unauthorized disclosure.
|Selecting the optimal LLM
|Ensure that LLMs can enable data protection, e.g., operate with Azure open AI if client data is on Azure, use open-source LLMs configurable within a client’s system, and encrypt proprietary or sensitive data before transmission to ensure security.
|Minimize AI-generated inaccuracies by designing the system to stay rooted in provided, verified data from specific documents.
Having made significant progress in addressing these challenges, the focus now shifts to the future possibilities of GenAI in banking, highlighting its potential to redefine the industry.
Where we’re heading…
As we look towards the future of banking technology, GenAI emerges as a game-changer in several key areas.
|Know-Your-Customer (KYC) Process
|Accelerates KYC by extracting and classifying information from documents, simplifying and speeding up verification.
|Generates synthetic data to balance incidence in fraud data, leading to lower false positives.
|Marketing and customer engagement
|Personalizes customer interactions and content, reducing fatigue; advanced chatbots enhance engagement with context-aware recommendations.
While these applications represent the potential of GenAI, it is essential to understand its current role and impact in banking.
…And where we are right now
The practical, current applications of GenAI in banking demonstrate how it is reshaping and streamlining processes and increasing efficiency in a cross-section of applications.
A critical application of GenAI is condensing information from extensive documentation into concise summaries. For example, GenAI can synthesize pivotal points from multiple lengthy documents into a compact, two-page summary. This is not just a simple extraction of excerpts but a sophisticated synthesis of information, conserving significant time that would otherwise be dedicated to reading and understanding each document individually.
When a bank considers extending a line of credit to a large corporation, GenAI can drastically reduce the time required for research and analysis. Typically, this task would demand extensive reading and synthesis by an analyst over several days. GenAI can automate the bulk of this process, allowing analysts to focus on the more nuanced aspects of decision-making rather than spending all their time on information gathering and initial analysis.
Handling unstructured data
GenAI proves highly efficient in managing queries related to unstructured data, such as those found in complex legal documents. For instance, front-end bank employees often need to access specific information from detailed, legally dense documents released by entities like the CFPB in the US. GenAI-powered chatbots can ingest these documents and provide instant, accurate responses to queries, such as information on pricing tiers or compliance issues. This saves time for the employees and ensures the accuracy and completeness of the information provided to customers.
Reducing compliance risks
The use of GenAI in extracting and summarizing relevant information from updated regulatory documents is crucial in reducing compliance risks. Given that many customer complaints arise from incomplete or inaccurate information, having up-to-date and precise data is essential. GenAI assists in keeping front-end staff informed and compliant with current regulations, an especially critical aspect in markets like the U.S., where compliance is strictly enforced.
Revolutionizing dashboard interactions
By enabling direct user queries for immediate, contextual answers, users can ask questions like, “What are the sales for the West region?” and receive specific data and contextual summaries. This innovation reduces the need for multiple dashboards and dependence on data analysis teams. GenAI’s dynamic response format, whether charts, tables, or summaries, streamlines data analysis and simplifies meeting preparations, as relevant, real-time data can be queried and displayed on the fly.
Optimizing contact center management
GenAI streamlines contact center operations by automating call categorization, conversation summarization, and sentiment analysis tasks. Traditionally limited by time constraints and resulting in incomplete data, GenAI now enables complete analysis of all call transcripts, overcoming previous sampling limitations. This thorough analysis enhances the understanding of customer interactions and trends. The insights derived are compiled into an analytics dashboard, improving call effectiveness and identifying areas for improvement.
As with any technology, great power comes with great responsibility to deliver optimized solutions. Fractal’s goal is to do just that.
Fractal’s distinctive approach
With 23 years in the industry, we have nurtured a deep understanding of the banking sector. This experience encompasses working directly with banks and with payment clients who collaborate with banks. Our extensive engagement covers various geographical locations and banking verticals across a broad spectrum of banking clients.
This breadth of experience provides Fractal with a comprehensive view of the banking industry, enabling it to identify where GenAI can have the most significant impact. This insight led to Fractal’s strategic advice against rushing to implement generic chatbots for bank websites, advocating instead for more impactful applications of GenAI.
Leveraging insights across various domains
Fractal’s diverse expertise in sectors like consumer-packaged goods, health, and technology equips it to innovate in the banking industry. For example, we transformed a CPG dashboard for banking analytics and decision-making and reconfigured call center solutions to improve banking customer service. Our product, FlyFish, initially tailored for e-commerce, now assists bank customers in product selection. This approach not only speeds up the adoption of new technologies in the traditionally slow-to-change banking sector but also elevates customer experiences and operational efficiency.
Roadmap for the next five years
The shift from initial hesitation to growing comfort with LLMs demonstrates GenAI’s deeper integration into banking operations — and its adoption is expected to accelerate, albeit in line with emerging regulatory guidelines for data privacy and consumer protection. While GenAI represents a significant technological leap, its application must be judicious and value-driven, enhancing customer experiences and operational efficiency without overcomplicating solutions. In the future, the banking sector will have GenAI evolve into a flexible, multi-LLM system tailored to specific needs and regulatory requirements, ushering in an era of improved efficiency, customer satisfaction, and adherence to data protection standards.