Behind ttb: Building Humanized Digital Banking with AI
An in-depth look at ttb spark developing ttb touch, integrating AI Yindee and ttb enterprise to deliver seamless financial experiences.
Stock photo for illustration only, not from the actual event
- ttb spark drives Humanized Digital Banking by combining Digital, Human, and AI.
- Yindee AI assistant on ttb touch utilizes an Orchestrator system managing 8-9 specialized bots.
- The ttb enterprise platform connects apps, branches, and contact centers, saving over 120 million baht in software costs.
- ttb spark is structured as an in-house unit where business and tech teams collaborate from day one.
Managing finances in real life often involves complexities far beyond simple money transfers or checking account balances. Individual customers face distinct financial milestones, ranging from systematic savings to loan applications and investments, all requiring reliable guidance tailored to their life stages.
Techsauce recently attended the Humanized Digital Banking event hosted by tth, where Naris Arakskulwong, Head of Corporate Strategy and Digital at ttb, shared the engineering background behind ttb spark. He emphasized that the primary challenge today is not merely technological advancement, but empowering customers to make sound financial decisions throughout their lives.
This vision led to the Humanized Digital Banking concept, which develops Digital, Human, and AI capabilities in tandem. The approach begins by deeply understanding customer pain points and leveraging data and AI to assist both clients and frontline staff in managing finances at the right moment.
Currently, the ttb touch application records between 1.5 to 2 million daily logins. To handle diverse customer requests efficiently, ttb introduced Yindee, an intelligent 24/7 assistant designed to resolve basic inquiries and reduce the time spent searching for information or repeating background details.
Stock photo for illustration only, not from the actual event
Technically, Yindee relies on an advanced Orchestrator Bot architecture rather than a standard single Large Language Model. This orchestrator analyzes customer context, intent, and sentiment before routing tasks to 8 to 9 specialized sub-bots handling credit cards, auto loans, mortgage loans, mutual funds, and insurance, while cross-referencing over 1,200 real-time promotions and product fact sheets to mitigate hallucination risks.
Implementing an orchestrator-based multi-bot system represents a crucial evolution for modern financial institutions. Because standard LLMs carry inherent risks of hallucination, delegating specific product domains to dedicated sub-bots significantly enhances the accuracy required for handling sensitive financial data.
Furthermore, Yindee integrates seamlessly with ttb enterprise, a proprietary platform built for frontline employees. Yindee currently achieves a 73% autonomous resolution rate with an 89% CSAT score, while complex cases are smoothly handed over to human agents equipped with full chat histories to eliminate redundant explanations.
Source: Techsauce
Found something wrong in this article? Report an issue with this article
Comments
Leave a Comment