By Jerry Jimenez Bongco, Country Leader, Philippines, AWS
Generative artificial intelligence (AI) offers something far more precious than monetary returns—it gives us the most valuable commodity of all: time. In financial services, this remarkable and rapidly evolving technology helps financial analysts, financial advisors, loan officers and others to remove the heavy lifting from time-intensive manual tasks so they can use their talents to think creatively and explore new initiatives faster.
Asia’s financial institutions are rapidly adopting generative AI and is set to grow at an eye-watering compounded annual growth rate of 96.7% by 2027. And as institutions across the region move from experimenting with generative AI to fully deploying it, it is critical that businesses “productionize” their proofs-of-concept to save time and accelerate innovation.
Here are some of the top trends in the FSI space:
Optimized for Every Use Case
Customers need access to a variety of Large Language Models (LLMs) to discover what works best based on their needs in the most time-efficient manner. AI is like a recipe, with blended ingredients like large logic functions and data sources carefully combined to produce a bespoke dish. With so many options in the AI kitchen, organizations can customize their AI systems to suit different needs. The companies that do this successfully have a recipe that comprises a thoughtful data management strategy with the responsible use of AI. This includes content moderation that detects biases in data and balances human and AI judgment.
Very much like a versatile kitchen that can craft a variety of dishes to satisfy different tastes and occasions, services like Amazon Bedrock are the ultimate solution for AI developers. Instead of having to source and prepare all the ingredients themselves, services like this provide a fully-stocked pantry of pre-made, versatile foundation models trained on a diverse array of data for multiple use cases.
Take U.S. investment management firm Bridgewater Associates, for example. The organization is using Anthropic’s Claude model to create a secure LLM-powered Investment Analyst Assistant that can generate elaborate charts, compute financial indicators, and create summaries of the results.
Meanwhile, the New York Stock Exchange is experimenting with different foundation models to automate tasks, better understand market sentiment, and gain valuable predictive insights into stock price fluctuations.
The exchange is also developing a Trading Rules Document Intelligence Chatbot using generative AI. This tool processes about 20,000 pages of trading rules from all U.S. exchanges, making it easier for users to ask questions and understand complex trading regulations.
Driving Agility and Compliance
Highly regulated companies may find it easier to adopt generative AI, enabling them to outpace smaller, less closely governed firms. Financial institutions, for example, possess extensive data from market research, trades, and financial information providers.
Generative AI can rapidly analyze this data to uncover valuable insights that drive success. While understanding and protecting this data is crucial, regulated companies already have established experience in safeguarding sensitive customer and financial data, making AI adoption faster.
This expertise in data management gives financial institutions a significant advantage over organizations that have not yet addressed these issues.
Empowering Humans Through Augmentation
The success of generative AI hinges on the sophistication of algorithms as well as the cultivation of a skilled workforce able to harness its full potential.
A team well-versed in generative AI can quickly iterate and fine-tune models, reducing the time and resources required to achieve desired outcomes.
According to the 2024 Asia-Pacific (APAC) AI skills survey commissioned by AWS , 95% of surveyed financial services employers in APAC expect to use AI solutions and tools by 2028, and this same group is also willing to pay a salary premium of over 40% for workers with AI skills.
BBVA, a global banking leader, has over 1,000 of its data scientists leveraging advanced skills using AWS’s generative AI services, enabling them to build and deploy machine learning models for a diverse range of use cases. Mitsubishi UFJ Financial Group (MUFG), Japan’s largest financial services provider, also plans to lean into generative AI to improve productivity across all lines of its business, including customer service, finance, human resources, and sales.
Today, for example, when analysts in financial services receive automated alerts of suspicious activity, they must conduct an initial review of whether the activity warrants further investigation.
This wasn’t the case for Nasdaq.
The stock exchange streamlined the review of suspicious activity alerts with AI to automate tasks like distilling information, analyzing filings, and summarizing news and market sentiment. In fact, during proof-of-concept testing, surveillance analysts gained an estimated 33% reduction in investigation time.
The Enduring Relevance of Traditional AI and ML
While generative AI represents a departure from traditional automation, it does not render traditional AI and machine learning (ML) obsolete.
On the contrary, it retains its relevance to address different needs.
The world of finance often involves risk modeling, portfolio optimization, and other quantitative tasks that require precise numerical processing capabilities. These tasks are better suited for traditional AI algorithms specifically designed for numerical analysis and mathematical operations.
The Generative AI Roadmap for Financial Services
The ability of financial institutions to “productionize” their proofs-of-concept comes down to a few things.
Firstly, it rests on the quality of their data and the robustness of their testing protocols. Just as a skilled chef requires the freshest ingredients to create an exquisite culinary experience, advanced AI models perform best when trained on high-quality, unbiased data.
Next, before releasing AI models into production environments, rigorous testing in sandboxed settings is paramount. Firms must “measure twice, cut once” by simulating real-world use cases and stress-testing their models against hypothetical challenges. Only then can they credibly lean into multi-model generative AI strategies that combine the strengths of different foundation models.
Crucially, creating an environment in which employees feel safe to experiment with AI tools without fear of failure, while developing clear career pathways for AI-focused roles, will be key to cultivating the human capital needed to propel this transformation.
The financial institutions that clear these bars will be rewarded with something that is priceless: time itself.
