Where AI is working
While the possibilities powered by AI are limitless, there are areas where AI is reported to have high-impact applications with tangible results already seen by financial institutions worldwide. In banking, loan processing shows high transformation potential with AI streamlining the process and cutting approval times from days to hours. AI accelerates the procedure by automating several steps of the entire process, from document checks to decision-making. Optical Character Recognition, a subfield of AI, can be used to scan and extract information from applications and submitted documents, saving inspectors time of manual data collection while reducing the risks of misinformation and human errors. AI algorithms can also apply pre-defined rules and predictive models to approve or reject applications automatically and suggest loan options based on the applicants’ profiles. For instance, a top international bank has co-developed an AI-powered digital lending platform with FPT and managed to accelerate its loan and credit application processing time from 6 days to under 24 hours. The platform is equipped with a virtual assistant to offer an easy and paperless credit application process, by providing product information, comparing different products, and recommending suitable options based on customer preference.
Customer service is another highly sought use case. AI chatbots can handle routine queries, such as balance checks and transfer requests in banking, or policy review and claims support in insurance. Unlike traditional chatbots that can only resolve a few simple questions with structured answers, AI virtual assistants can generate human-like interactions and complete a large number of complicated tasks. This automation of routine queries not only improves operational efficiency with faster services and higher accuracy, but it also enhances customer satisfaction with shortened processing time and 24/7 operations while freeing human agents for more challenging tasks. For example, an Asian insurance group has partnered with FPT to roll out a super app for its customers. The application features a range of self-services and an integrated AI chatbot for functions such as policy review, claims submission, and even health monitoring, allowing the insurer to resolve a whopping 87% of customer queries in the first call.
In financial services, portfolio management has experienced significant improvements by applying AI. With the ability to analyze a wealth of complex and unstructured data, AI possesses the ability to identify patterns that are often invisible to the naked eye. By flagging non-obvious correlations and anomalies, AI is able to surface early signals and adjust portfolio management strategies accordingly, even showing superior performance compared to human analysts. Indeed, researchers at Stanford University created an AI analyst to assess and adjust fund managers’ portfolios using nothing more than public information. The model first developed a stock-picking acumen from its training input fed with market data between 1980 and 1990, which was then used to correlate 170 variables with future stock performance. Researchers then asked the AI model to assess and improve roughly 3,300 portfolios actively managed by fund managers between 1990 and 2020, and adjust them once per quarter. At the end of the simulation, the AI model managed to generate $17.1 million per quarter of alpha, while in reality, between 1990 and 2020, fund managers generated $2.8 million per quarter. This leads to a conclusion that AI outperforms 93% of fund managers in alpha generation by an impressive rate of 600% on average [2].
In insurance, AI is expected to facilitate a complete transformation of the claims process, with four out of five executives showing confidence in technologies such as automation, AI, and ML-based data analytics [3]. By streamlining the entire claims process, from assisting with application review and decision-making to fraud detection, AI is the key enabler for driving a digital, seamless, hassle-free claims experience. OCR technology can automatically scan claims applications and submitted documents, extracting said information and uploading it to the processing system with minimal errors while speeding up the process. AI also assists decision-making by applying pre-defined rules to approve and reject claims requests, ensuring decision consistency and accuracy, which the traditional human-based approach lacks due to the heavy reliance on assessors’ experience and expertise. AI algorithms also help flag anomalies and detect fraudulent claims, which is challenging for human inspectors processing a massive amount of claims data. Subsequently, applying AI in claims processing not only streamlines the process and improves decision consistency but also reduces fraud risks and ultimately saves costs for insurers. A real-life example is an insurance group employing FPT’s AI-powered solution – Insurance 360 – to automate the processing of 140,000+ claims requests annually. The insurer managed to cut claims processing time from 2 days to 2 minutes and save 8% in costs by detecting fraudulent claims.
How to make it work?
As AI matures, the winners will be those who embed it thoughtfully, combining speed, trust, and domain understanding. Through FPT’s experience in accompanying global financial institutions on their AI journey, four lessons can be drawn to turn AI visions into tangible impacts:
1. AI works best when tied to business needs—not hype:
The limitless possibility offered by AI is twofaced. While it provides an exciting future where AI enables unimaginable experiences, financial organization leaders are left puzzled and directionless about how and where to deploy it effectively. A survey by Ernst & Young LLP reveals that while a whopping 99% of financial services leaders reported that their organizations were deploying AI, 20% of the leaders lack confidence in their ability to capture the potential of this technology [3]. The challenges often stem from a range of barriers, such as the absence of a clear digital transformation roadmap, outdated legacy systems, and inadequate data and technology infrastructure. To unlock AI’s full potential, financial institutions must start with a solid foundation: aligning their AI strategy closely with their broader digital transformation goals and core business needs.
2. Change management and training are critical:
Fully embedding AI in the corporate culture and day-to-day activities requires more than just technology implementation; it involves a complete transformation of employee skills and mindset. On top of the extensive training required to effectively help employees confidently use AI at work, resistance proves to be the biggest challenge. With a whopping 75% of employees fearing that AI could eliminate certain jobs [4], resistance to change is both expected and significant. Developing a strong change management strategy with defined goals, clear implementation roadmaps, and effective and open communication is, hence, critical.
3. One-size-fits-all doesn’t work—context matters:
Given the diversity in regulatory landscapes, customer segments, and technological capabilities, financial institutions must design AI strategies that are tailored to their specific business contexts. This need for customization extends even within multinational organizations, where overseas branches must adapt AI implementation plans to fit local market dynamics and compliance requirements. For instance, AI-driven credit scoring models must account for the differences in data privacy regulations—while the EU strictly limits the use of non-essential personal data, the US permits broader data usage. These regulatory difference have prompted global banks like HSBC and Bank of America to develop “region-specific AI variants,” at a cost 40–60% higher than deploying a standardized model [5].
4. Explainability is non-negotiable in regulated environments:
With 63% of UK banks reporting difficulties in meeting explainability standards without sacrificing model performance [5], explainability has emerged as one of the most pressing challenges in AI implementation. This issue is further compounded by the rapidly evolving regulatory landscape—where regulatory changes have surged by 500% since 2008 [6]. This growing complexity makes compliance more challenging than ever before, meaning that financial institutions must keep a close eye on their AI solutions to keep them strictly updated with the changing regulations.
