Ant International is attempting to modernize real-time treasury operations by integrating advanced generative AI into Hong Kong’s regulatory testing framework. Through its Ant Bank and Bettr subsidiaries, the company has joined the GenA.I. Sandbox++ to test AI-driven liquidity risk management. This move seeks to align digital banking capabilities with the continuous, 24/7 nature of modern digital financial activities and cross-border payment flows.
Ant Bank and Bettr Deploy Falcon TST 2.0
The initiative focuses on deploying the Falcon Time-Series Transformer (TST) AI Model 2.0 to enhance foreign exchange and liquidity management. Ant Bank intends to use this proprietary model to generate granular, statistically-grounded daily forecasts, aiming for more proactive liquidity planning within its treasury management functions. Bettr, acting as the technology partner, will utilize its Platform Tech to facilitate the deployment of the Falcon TST 2.0 model for the bank's treasury team. This technical integration is designed to support a 24/7 financial ecosystem, addressing the specific cashflow prediction needs of customers who engage in constant digital financial activities. The company is positioning this deployment as a method to establish new best practices for AI-driven liquidity risk management within a regulated environment.
Regulatory Integration via GenA.I. Sandbox++
The participation occurs within the GenA.I. Sandbox++, a collaborative initiative launched by the Hong Kong Monetary Authority (HKMA) alongside the Securities and Futures Commission (SFC), Insurance Authority (IA), Mandatory Provident Fund Schemes Authority (MPFA), and Cyberport. This sandbox is designed to foster AI innovation across Hong Kong’s financial industry. Ant International’s involvement highlights its strategy to collaborate with the HKMA to develop the local banking ecosystem. A key technical benchmark for the project is the Falcon TST 2.0 model, which the company claims achieved state-of-the-art performance on the Mean Absolute Scaled Error (MASE) metric. According to the company, the model has demonstrated forecast accuracy exceeding 93% on a leading global public evaluation benchmark for time-series foundational models.
Key Takeaways
- Ant Bank and Bettr have joined the HKMA-led GenA.I. Sandbox++ to focus on liquidity risk management.
- The project utilizes the Falcon Time-Series Transformer (TST) AI Model 2.0 to improve FX and liquidity forecasting.
- Ant International reports that the Falcon TST 2.0 model has achieved forecast accuracy exceeding 93% on global benchmarks.
FinanceInsyte's Take
In our view, Ant International is testing whether high-accuracy time-series models can effectively bridge the gap between traditional banking hours and the relentless demands of 24/7 digital commerce. By moving Falcon TST 2.0 into a regulatory sandbox, the company is not just testing a product, but is attempting to validate AI-driven treasury management as a standard for digital-first banks. If successful, this could signal a shift where liquidity management moves from reactive, human-led processes to proactive, model-driven continuous monitoring.
Questions & Answers
How does the Falcon TST 2.0 model impact treasury operations?
The model is intended to provide granular, statistically-grounded daily forecasts to support more precise and proactive liquidity planning for real-time treasury management.
Which regulatory bodies are overseeing the GenA.I. Sandbox++?
The sandbox was launched by the Hong Kong Monetary Authority (HKMA) in conjunction with the SFC, IA, MPFA, and Cyberport.
What specific technical metric is cited for the Falcon model's performance?
The company states that the Falcon TST 2.0 achieved state-of-the-art performance on the Mean Absolute Scaled Error (MASE) metric.
What is the primary strategic objective for Ant Bank in this sandbox?
Ant Bank aims to leverage AI and LLM technology to improve cashflow prediction accuracy to better reflect the 24/7 digital lifestyle of its customers.
Source: Businesswire
