Quick-Win AI Strategies for Banks

Over the last two years, there has been significant buzz around the potential of GenAI and the various platforms and tools that have emerged in the market. However, we have not seen the same level of buzz around successful implementations, barring some examples of incremental progress. Only 2% of organizations in financial services are considered “leaders” in AI and analytics according to a 2024 assessment by global consultancy Kearney.

This series would focus on some key principles that can help accelerate the implementation of AI. We encourage your comments and suggestions, and hope to build this series into a valuable reference source. 

Principle 1 – Extending Existing Capabilities 

Context 

When banking leaders talk about their AI journey over the last 15 years, it often revolves around two components: Natural Language Processing (NLP), and Machine Learning (ML). 

NLP applications have primarily focused on: 

• Converting voice to text 
• Processing text in various formats
• Converting text to voice (in limited applications)

ML applications have concentrated on:

• Pattern identification (trends and cohorts) 
• Forecasting

Scenario

Financial institutions that have ventured down these paths would have likely:

• Built custom platforms using open-source tools or limited commercial ones
• Invested time in training these platforms using their existing data to fine-tune performance and achieve production-grade accuracy
• Primarily used structured data as the source for NLP-based searches
• Implemented periodic recalibration of ML models based on application requirements 

Key Considerations: 

When considering AI enhancement, examine these factors 

• The potential value of incorporating unstructured data - can the existing platform support it?
• Whether new features can enhance efficiency? [For example: semantic versus lexicon search, or transitioning from ML to Deep Learning (DL)]
• The role of GenAI in reducing manual intervention – can it integrate seamlessly with the existing platform? 
• Whether there are sufficient upgrades to the current stack – would moving to a newer stack with accelerated enhancements improve performance?

• The sustainability of the platform – does it require specialized resources, or can it be managed effectively by business teams with minimal technical expertise?

Strategic Options

Banks are left with a decision: 

• Enhance the current platform by integrating with new-age tools and platforms
• Make component-level transformations, while maintaining the core architecture
• Migrate the entire technology stack to a new, more advanced stack

Conclusion

The case for AI is clear. McKinsey’ Global Institute forecasts that GenAI could add between $200-$340 billion in value annually for the global banking sector, largely through increased productivity, while a Bain & Company 2024 survey of U.S. financial firms found an average productivity gain of 20% across uses from GenAI. 

As banks navigate the evolving AI landscape, from GenAI to Agentic AI, the key to success lies in making informed decisions based on existing capabilities and future needs. Rather than pursuing complete overhauls, organizations may need to consider building upon their existing capabilities, keeping in mind their technology infrastructure’s scope and ensuring regulatory adherence, to maximize the value of AI. 

We look forward to hearing your perspectives and experiences on AI!