The AI automation implemented in the financial services industry largely takes form of Robotic Process Automation (RPA), followed by Virtual Assistants such as chatbots, and Machine Learning (ML). It is estimated that ML adoption among banks will continue to skyrocket in the near future, as more and more banks are implementing it across the market to assist them in fraud, underwriting and risk management.
When it comes to AI adoption, there are two main factors involved; the AI technology itself and the adoption. Under AI, banks can choose from three commonly used technologies which are ML, Computer Vision and Natural Language Processing (NLP). ML covers algorithms and statistical analysis, as it has the capability to study the data gained in order to provide prescriptive and predictive analyses; hence coming up with smart and personalized recommendations based on the data outcome.
Whereas Computer Vision is leveraged in image processing, as it can understand and differentiate images. It is intelligent enough to drive certain understandings from the images uploaded. Lastly, for NLP, it can help banks to understand human language without the need of human labor. It has the capacity to supplement and derive contextual analyses, while understanding customer concerns and respond to them accordingly.
When it comes to implementation, banks then shift to the automation, which consists of RPA, Intelligent Process Automation (RPA), and Embedded AI. RPA is commonly utilized to automate rule-based, repetitive tasks and workflows. Banks usually encourage their employees across branches and business units to implement this in order to optimize AI automation, hence reducing the need of manpower usage for manual, administrative tasks. Common processes that are often assisted with RPA include; internal operations, payrolls, reports consolidation, report automation, KYC, account closure, credit card processing and loan processing.
As for IPA, it is often paired with RPA as it can add in cognitive intelligence. IPA can augment human processes more effectively, as RPA alone is not sufficient to derive human-alike decisions. This is as RPA is not able to diagnose and understand past historical data to make human decisions, whereas IPA is equipped to do so, making it the more popular choice among banks.
With embedded AI, banks can enjoy AI built-in platforms, as many vendors already optimize AI in the applications that they offer. This makes it easier for banks and FIs to implement new solutions with platforms that already pre-optimized with AI. Where RPA and IPA focusing into existing manual processes and workflows in heterogeneous environment within existing legacy systems, embedded AI is much more efficient as it comes with the new platforms that adopted by the bank.
Notable AI Case Studies 1. eKYC Customer Onboarding
With the pandemic accelerating digitalization across all industries, same goes with the financial landscape. One of the most prominent improvement that banks have achieved is the existence of eKYC, which allows remote onboarding process. This allows banks to forego branches and physical meet-ups for account openings, as all the new customers need is their designated mobile application or website.
2. From OCR to cognitive data capture
Another major breakthrough in the industry, Optical Character Recognition (OCR), can blindly read a scanned document into a system – without knowing the actual meaning of the data. This has fortunately evolved into cognitive data capture, in which the technology is intelligent enough to recognize and understand the context of document and content. This undoubtedly improves the bank’s turnaround time and productivity as it reduces the need to invest in human labor.
3. Anti Money Laundering (AML)
Not adhering to the AML compliances can cost banks up to $574 million, which is no small amount. Hence, banks tend to be extremely careful in adhering to the AML rules and regulations. AI powered AML solutions is often utilized by banks as the technology uses ML to optimize the existing rule-based detection engine. It has the capacity to predict the likelihood of suspicious transactions by classifying them into a few categories. AI-powered AML solutions often can identify transactions which may be missed by the typical rule-based detection engine; hence able to detect any new pattern of transactions that could be suspicious. It comes with higher accuracy and productivity, ensuring compliance at all times.
4) Credit decision engine
The AI-powered credit decision engine can assist banks in strengthening their credit management capabilities. Many banks have started to build an AI-model credit scorecard which are using their existing data to optimize storage. Data from several years back are analyzed and customer demographics are compared against new applicants. This is to effectively prevent credit loss to new applicants who do not have previous credit history with the banks. Said engine also can improve the accuracy of predicting the person’s credit score, which augment the process to auto approve or auto reject applications accordingly.
Cognitive data capture – Using AI to Understand and Act on Content and Documents
AI is incredibly helpful for banks to contextualize documents, which often comes in three forms; structured, semi-structured and unstructured. Structured data typically comes in the form of identification cards, and so forth, in which the content and details can be easily identified. Semi-structured data consists of invoice, orders, PO and delivery notes, which ted to have different layouts based on the suppliers. Lastly, unstructured data encompasses of emails, newsfeed, and social media – in which banks can struggle in deriving the details despite understanding the context.
The challenge is as follows, it is predicted that by 2025, up to 80% financial data will be unstructured. This poses great difficulty for banks to extract the right information, which leads to them needing intelligence and AI to function appropriately. OCR is the first step, as it scans images and take out information as needed. There are other extraction methods that can be seen below.
Moving forward with the right AI strategies
AI strategy across banking channels differs, as it follows the structure of front, middle and back offices. For front office banks tend to focus on customer experience; typically implementing chatbox, voice box, customer support. Efforts are directed mainly to improve the personal financial management offered to the customers, ensuring a hyper-personalized customer experience. This is to improve acquisition and satisfaction rates, therefore reducing costs.
In the middle office, banks are leveraging KYC to transform themselves in becoming fully automated. Topped with AML, they are more than equipped to reduce risks, costs and time. Whereas in the back office, credit underwriting is usually implemented. This is as it can majorly benefit the banks that can leverage on AI as their credit risks, human error, labor time are significantly reduced.
Transform your bank digitally with SYNERGi
Being a trustworthy technology enabler, INFOPRO’s SYNERGi is guaranteed to help you realize the four aspects of AI automation; RPA/IPA, Embedded AI, AI Service Hub and AI Talent Pool. INFOPRO can seamlessly automate the heterogenous environment found in legacy systems with minimal customization needed with RPA/IPA. Whereas we also offer AI-embedded in all our applications, as system capabilities are already built in across all apps we developed.
SYNERGi also offers centralized AI model repository to train, build and deploy your AI models as API that can be utilized by any of your applications across all business sectors and systems. This as it takes the form of a neutral diagnostic, available in the form of AI. Lastly, INFOPRO’s expert talent pool can help you to build custom AI models and resolves problems within your organizations.
Interested to know more?
Contact us for more information.