Image generated by Gemini
Given the rapid changes in customer behavior, incumbent financial institutions must adapt to remain competitive in the market. Customer touchpoints and experiences become increasingly important factors in differentiating their position.
As digital banking has become increasingly common, customer expectations for timely support have understandably risen. However, financial institutions often struggle to achieve deep consumer engagement solely through mobile apps. For more complex products like investment funds, bonds, or intricate investment vehicles, human assistance remains crucial. Human advisors can guide customers, understand their risk appetite, and find suitable wealth products. Some customer requests cannot be done purely by clicking the button. It is important for financial institutions to understand the “situation” and “unique requirement” customers are demanding, while being attuned to the customer emotional state, and response accordingly. Since customers are active on various channels, effective communication and engagement method are vital in capturing their attention and meeting their expectations.
Therefore, banks must consider leveraging new technology and automation, on top of already existing tools, to boost customer satisfaction while streamlining operational costs. One promising solution is “conversational banking.”
Conversational Banking: An Evolution of AI for Financial Services
Conversational banking is a rapidly growing trend in the financial services industry, improving the way banks interact with customers by leveraging technologies such as artificial intelligence (AI), in conjunction with existing chatbot, to deliver more personalized and accessible banking experiences.
Customers nowadays often expect faster response time from their preferred communication channels. Chatbots are among the most common applications to expand banks’ service time to 24/7 and the earliest applications of conversational AI in banking. They are on the edge to help customers from mundane financial activities such as transferring money and checking account balances, to less straight-forward activities such as account opening, managing investment portfolio, and negotiating credit card payment terms, all without the need to tediously scan through websites or apps or wait on hold for a call center.
However, due to their constraints around the ability to comprehend only specific use cases and precise keywords, chatbots often force customers to constantly adapt their language to structured commands or predefined phrases, which can be frustrating for customers and loss of opportunity for banks. Based on this pain point, conversational banking offers a more natural ways for customers to interact with banks, and the engine adapts over time using machine learning to learn from past interactions and make the chatbots smarter.
In short, conversational banking takes traditional chatbots to the next level. It empowers chatbots to learn from past interactions and anticipate user needs, while able to understand natural language used by customers, resulting in a more engaging experience through “human-like” interactions. These “smarter chatbots” leverage various technological tools to enhance their capabilities and achieve more for both customers and banks.
The Tech: Automation of Human-like Interactions from Rich Data
Unlike static chatbots, conversational AI learns from every interaction with its users. Each conversation feeds its machine learning (ML) engine, enabling it to handle advanced terms, local slang, and dialects. Behind conversational AI, there are various supporting technological tools to make the bot learn, analyze and response.
In a non-technical term, the AI first tries to understand what a user is asking, then chooses the best way to respond, and finally makes sure it sounds natural. If the user is using voice, it also needs to understand what the user says, both message and tone, and replies clearly and empathetically.
As illustrated above, conversational banking leverages a range of technologies to resolve customer queries including Natural Language Understanding (NLU), Dialog Management, Natural Language Generation (NLG), and Automatic Speed Recognition (ASR) (details of each tech are noted in the table below). Each tool works together to make the bot function. NLU and NLG are parts of Natural Language Processing (NLP).
NLP offers a range of powerful capabilities, including speech recognition, speech-to-text conversion, and text-to-speech synthesis. Furthermore, NLP can now even identify emotions in text or speech, adding a new level of understanding to human-computer interaction.
Importantly, the machine learning tool helps the bot analyze and learn from past conversations, and applies them to the conversation at hand.
Table 1: Bundled AI tools help the bot work and learn.
| Functions | Tools |
| Understand the intend behind a text | Natural Language Understanding (NLU), a part of NLP |
| Form a response | Dialog Management |
| Generates a response in a human-friendly manner | Natural Language Generation (NLG), also a part of NLP |
| Convert speech to text and text to speech | Automatic Speed Recognition (ASR) |
| Learn from experience | Machine Learning (ML) |
Due to technology like NLP and ML, AI has become smarter and more human-like. Additionally, another key development driving this progress is the data used to train models. In the past, traditional single-model AI relied on a single source or type of data for specific tasks such as scribing texts from the internet to teach AI. However, multimodal AI ingests and processes data from multiple inputs like text, video, images, and speech. By combining relevant data from various sources (not limited to text), AI has vast amounts of information to learn and analyze, which could be put together into smarter and more engaging responses.
The Promise: What Human-Like Chatbots May Bring to the Market
The way financial institutions interact with customers is constantly evolving. Conversational banking disrupts these traditional methods by leveraging new technology to respond to ever-changing customer behavior. For the purpose of this article, we can examine the impact of technological advancement on how banks interact with customers.
Tech Advancement: From Static to Dynamic Scripting
As mentioned earlier on chatbots vs conversational AI, the technology advances from bots responding to simple questions or requests, based on pre-analyzed models and logic, to adding some sorts of analytics and personalization. Today, due to reduced technology costs of AI development and cloud computing, real-time analytics is made possible. Bots could gather data from various sources in real-time to analyze and predict customers’ next requests.
The form of conversational AI is enabling banks to move away from making simple requests like transferring money, to provide deeper analysis such as spending graphs with monthly comparisons, to personalized solutions for each user, and real-time monitoring to report and support. Now the next move is Anticipative Interaction. AI anticipates customers with specific needs before they even reach out. Even before customers get in touch, an AI-supported system can anticipate their likely needs and generate prompts for the agent (as co-pilots) or for decision maker to fabricate pre-made solutions. For example, the system might flag that the customer’s credit-card bill is higher than usual, while also highlighting minimum-balance requirements and suggesting payment-plan options to offer. If the customer calls, the agent can not only address an immediate question, but also offer support that deepens the relationship and potentially avoids an additional call from the customer later.
Interactions: From Reactive to Anticipative
- Where or Channels
Customers are migrated from traditional branches to digital channels. Traditionally, the main interaction between banks and customers is between bank branches and ATMs. Today, mobile banking is becoming a main point of contact, therefore, call centers and chatbots are playing larger roles in customer support.
Soon, chatbots and conversational AI will expand to every touchpoint of clients. The service could be integrated with different apps (rather than mobile banking apps) and channels (even branches to lower headcount costs) that users are interacting with. One example is the accessibility through third-party messaging services and social media platforms to improve the experience to customers. A customer may see a news recap on political conflict in a neighboring country on social media, forward it to the bank’s official account chat-box, and inquire about next steps on mitigating the effect on his/her investment portfolio. Customers would benefit from 24/7 support, reduced waiting time, and more personalized responses.
- What or Use cases
In the past, banks primarily used conversational AI for customer verification (Know Your Customer or KYC) and handling simple transactions like balance checks and money transfers. However, as advanced conversational AI can learn from experience, it opens doors for new use cases in the financial service industry.
The use of data can supercharge the bots’ analysis, providing deeper customer insights and fostering broader financial service integration. For example, conversational banking could support pre-qualification for loans by analyzing non-financial data such as voice to predict credit scores. It could also remind customers about upcoming payments based on their spending patterns or approve loan extensions using past repayment and social media data.
Looking to the future, conversational banking is expected to extend beyond banking products and services. By connecting data with third-party platforms, it could help customers improve their quality of life. Financial institutions could partner with other service providers such as exchanging bank loyalty points for airport pickups, or automatically ordering groceries for home delivery, or identifying surges in utility bills and then suggesting a financing plan for solar rooftop purchase. These services could be learned from past transactions and predicted by AI.
- How or Humanness
Conversational banking evolves chatbots beyond simple FAQ responses. Initially, they are trained with pre-defined prompts and answers. As they progress, they learn industry-specific vocabulary and become more flexible in understanding customer keywords to find relevant information.
Today, the level of analytics has greatly improved. AI will be built and currently have built to be more like humans with the learning of emotions and tone of voice when dealing with customers. Additionally, as the AI learns through multiple conversations, it increases the level of personalization and ability to support individual clients.
For example, in various use cases like cross-selling new products and collecting loan payments, AI will identify the right tone of voice such as softer and higher pitch voice to interact with each customer. Adversely, AI would analyze the emotions of customers during the interaction to identify suitable responses or when to escalate the conversation to the human support team. To drive a personalized experience, servicing channels are supported by AI-powered decision-making, including speech and sentiment analytics to enable automated intent recognition and resolution.
Why Now?
AI adoption poses benefits in terms of cost reduction, improved customer satisfaction, and increased competitive advantage. To stay competitive, embracing AI is no longer a choice, but a necessity. Leading institutions are already leveraging advanced AI to serve customers, empower employees, and secure their market share.
Higher automation reduces costs and improves customers’ satisfaction through operational efficiency, minimizing errors and optimizing resource utilization. Conversational banking further enhances customer experience by streamlining human resource allocation, reducing response times, and improving account access and security, while also reducing fraud. Conversational AI plays a key role by integrating data from various teams. This comprehensive data view empowers institutions to manage resources more effectively and ensure compliance with the evolving market and regulatory landscape.
HSBC’s use of conversational banking serves as a great example. Launched in June 2018, HSBC’s AiDA chatbot is used to respond to clients’ requests via instant messaging, reducing the cost related to calls by 90%. In February 2021, HSBC used a chatbot powered by AI to provide instant pricing and analytics for foreign exchange options, making complex trading more accessible and efficient.
Leading financial institutions are increasing the use of advanced AI technologies. McKinsey’s Global AI Survey reveals that nearly 60% of financial-service sector respondents have already embedded at least one AI capability into their operations. This digital transformation is taking shape through:
- Conversational bots for basic servicing requests
- Humanoid robots in branches to serve customers
- Machine vision and natural-language processing to scan and process documents
- Machine learning to detect fraud patterns and cybersecurity attacks from conversations
Technology helps speed up the tedious process and push financial institutions ahead of other banks and Fintechs. The new tech adoption will become standard practice or baseline in the eyes of customers.
Consumers are increasingly shifting towards digital channels, favoring mobile banking for simple services, and reducing visits to physical branch. This behavior shifts emphasizes the need for banks to adapt. Mobile banking, pioneered by commercial banks in Thailand, has been a successful driver of the digital trend. Presently, over 80% of bank clients have and regularly use mobile banking applications. In addition, government initiatives such as PromptPay and the G-Wallet policy, aimed at boosting domestic spending, have further accelerated the adoption of digital payments and mobile banking. A 2022 Mastercard survey revealed that a staggering 94% of Thai consumers now use digital payments. With Gen AI gaining popularity in public eyes, commercial banking is expected to play a an increasingly significant role in supporting banks’ customers in the near future.
Non-banking businesses are entering the banking space. Banking business is now embedded in a wide range of software and applications (See more about Embedded Finance here). One significant threats to banks is the emergence of “Super Apps” (See more about Super Apps here). These Super Apps integrate various financial services, including payments, and in some cases, lending, and insurance, potentially becoming one of the main operating businesses and posing a threat to incumbent banks. They disrupt traditional methods of offering new banking products and services and may soon seek to expand their presence and involvement in financial services on a larger scale. As a result, financial institutions will need to reassess how they participate in digital ecosystems and leverage AI to unlock the untapped data potential for competitive advantage.
Things To Be Cracked
Image generated by Gemini
Conversational AI could help boost banks’ presence and support clients; however, the technology presents challenges and hidden costs. New tech adoption is poised to disrupt how banks manage and operate. Incumbent players face a balancing act: ensuring agility and flexibility for competition while maintaining security and compliance to secure trust as financial service providers.
- Infrastructure Readiness:
Implementing conversational banking requires robust computing power and flexibility to support real-time analysis. However, legacy core banking systems are often difficult to modify. Additionally, fragmented data across different teams hinders the analysis of relevant data and timely generation of recommendations. Most importantly, transitioning to a new infrastructure and ongoing computational requirements can incur significant costs.
- AI and Talent Management:
A successful AI strategy requires a clear roadmap for both technology and talent. Currently, a major challenge is the lack of standardization in conversational AI adoption. Organizations are still exploring the best ways to integrate technology. The strategy should encompass both the technical process of tech implementation (building, testing, deploying, and monitoring) and talent development to uplift existing employees to develop and maintain new products and services. Ultimately, the strategy must identify and develop use cases where conversational AI can transform customer journeys, leading to defined outcomes such as real-time client support, tailored service with insightful data, improved customer lifetime value, and lowered operating costs.
- Regulatory and Ethical Considerations:
Disruptive technologies often raise regulatory concerns, particularly for financial institutions where reputation and stability are critical. This is especially true for conversational AI, which handles sensitive personal and financial information. Personal data management and changing regulations should be closely monitored when adopting conversational AI. Additionally, there are ongoing developments in the framework to help reduce human biases (such as social profiling) imposed on AI models. For example, Thailand’s Electronic Transaction Development Agency (ETDA) has published AI Governance Guidelines with hopes of bolstering domestic ethical AI development and adoption.
Closing Thoughts
Conversational AI offers numerous benefits in enhancing the customer journey. However, as with any technology, it comes with associated costs. Financial institutions should be aware of the limitations of AI technology and carefully balance the costs and benefits of adopting it. Ultimately, the success lies in the hands of banks that could identify where best to implement new technology and where its implementation might not be the most suitable solution.
Author: Panuchanad Phunkitjakran (Pook)
Editors: Supamas Bunmee (Jae), Woraphot Kingkawkantong (Ping)