AI in Financial Services: creating value from textual data

Despite its critical role, financial institutions only statistically model around 10% of it. That’s because 90% of financial institutions’ data exists as text, such as emails and documents. This type of data historically has not been analysed systematically or at scale using computers. As a result, financial institutions have missed out on the ability to use this data to inform critical strategic decisions and drive efficiency in core operations.

However, with the advance of Generative AI, and, more specifically, recent progress in large language models (LLMs), now means computers can analyse and generate terabytes of text with ease.

Using text to re-engineer financial services

At Faculty, we are helping financial institutions make the most of their text data by leveraging advances in AI. For example, we have expanded analytical capabilities at a large asset manager by programmatically processing vast volumes of text relevant to investment decisions. 

Additionally, for a large insurer, we have automated the extraction of key information from email attachments so that applications and claims can be actioned quickly. Our team is also helping to transform a leading bank’s customer service to answer customer queries more efficiently and effectively, improving overall customer satisfaction. 

New risks on the horizon 

Applying AI to text has its challenges. The ‘creative’ responses of LLM-powered AI agents can cause problems when consistent answers to customer questions are desired. Also, cybercriminals are exploiting the ability to create text computationally with speed and at scale to launch increasingly sophisticated spear phishing attacks on financial institutions.

But perhaps the biggest risk of all is that many financial institutions we speak with are not thinking about how to integrate various AI applications into a coherent whole. At Faculty, we believe the future of large language models is the foundation for enterprise AI systems powering strategic decision-making. Text-to-task agents will link various models and modalities (speech, text, audio, and images) as inputs and outputs.  

If you’re building Gen AI applications using text and haven’t considered how to connect them, then you’re likely building deeper siloes that will negatively impact the overall performance of your business. 

Capturing value from AI: focus on what matters

So here are three simple suggestions to ensure you maximise the impact of your AI investments:

1. Focus on business goals, first and always
   If you’re developing an AI strategy separate from your business strategy, you’re guaranteed to waste a lot of time and money. Instead, start by thinking about the critical decisions you must make as a business and how a system that lets you look ahead could help you achieve your goals. This will focus your attention on where AI can meaningfully drive value. In financial services, this means eagle-eyed attention on P&L value drivers: increasing yield, diversifying income, driving down operational costs, and reducing risks.

2. Focus on people
   Second, your AI should be human-centred. It should be designed to support you—not take your place. Just in the same way that ATMs reshaped bank branch staff roles rather than replacing them, AI can remove mundane work to free people to focus on the strategic aspects of their roles. This requires AI to be built in specific ways, with interpretability, safety and governance embedded from the start.

3. Focus on long-term enterprise change
   Finally, think about AI solutions as a series of individually valuable but collectively transformative projects. There is no need to boil the ocean. Invest in valuable use cases, but build them on a common architecture so they can be connected over time. This will ensure you experience all the benefits of long-term transformation and deliver value along the way.

Investing in AI? Want to maximise its value? Contact our team today.