Barclays, Citi, Deutsche Bank, and Standard Chartered are now using Ant International's FalconTST 2.0 AI platform to improve forecasting and manage currency risk in cross-border payments, with new features designed to address real-world data challenges
Several of the world's largest banks have begun using FalconTST 2.0, an artificial intelligence-powered forecasting platform developed by Singapore's Ant International, to strengthen their management of currency risk and improve the accuracy of financial forecasts. The platform, which builds on time-series transformer (TST) technology, is now operational at Barclays, Citi, Deutsche Bank, and Standard Chartered, where it is being integrated into systems that support cross-border payments and foreign exchange (FX) exposure management.
FalconTST 2.0 is designed to help financial institutions and their business clients navigate the complexities of cash flow forecasting and currency fluctuation. One of its key features is the ability to distinguish between missing data-such as the absence of bank transactions during weekends-and genuine zero values, reducing the risk of misleading patterns in forecasts. The system also adapts to new business scenarios by learning common time-series patterns and can process data at multiple time frequencies, making it suitable for sectors with rapidly changing transaction volumes.
AI Integration in Bank FX Platforms
At Barclays and Citi, FalconTST 2.0 has been embedded within the banks' FX hedging platforms-BARX NetFX and Fixed FX Rates, respectively. This integration is intended to support risk management for clients operating in sectors like e-commerce and aviation, where exposure to currency movements can have a significant impact on costs and revenues. Standard Chartered, meanwhile, is using FalconTST 2.0 alongside its SCALE FX system as part of its participation in the Monetary Authority of Singapore's PathFin.ai programme, which aims to advance the use of artificial intelligence in financial services.
According to Ant International, the new platform delivers more accurate forecasting in real-world FX risk management, particularly for cross-border payments. The company claims that FalconTST 2.0 has consistently improved forecasting accuracy by at least 93% in operational settings, though the precise methodology and comparative benchmarks have not been independently published. The platform's deployment reflects a broader trend among global banks to leverage advanced analytics and machine learning to address the challenges of volatile exchange rates and increasingly complex international payment flows.
Data-Driven Risk Management and Industry Context
In recent years, banks and businesses have faced heightened currency risk due to unpredictable market conditions and shifting monetary policies. For example, the pound sterling (GBP) has experienced notable fluctuations against the US dollar (USD) and euro (EUR) since 2022, with daily spot rate swings occasionally exceeding 1%-a level of volatility that can materially affect cross-border settlements and hedging strategies. As banks seek to manage these risks, the adoption of AI-driven forecasting tools like FalconTST 2.0 is becoming more common, especially in sectors where transaction timing and exchange-rate exposure are critical.
Ant International's approach to innovation extends beyond AI forecasting. The company has also partnered with Bank of China (Hong Kong) to enhance cross-border payment capabilities for its subsidiaries, using technologies such as blockchain to streamline settlement and compliance processes. This reflects a wider industry movement toward integrating advanced digital infrastructure to support international business payments and reduce operational friction. For readers interested in how diverging monetary policies and market expectations can influence currency performance, a recent analysis of the Australian and New Zealand dollars explores these dynamics in detail: how yield differences are shaping the AUD/NZD exchange rate.
Understanding AI in Currency Forecasting
AI-based forecasting platforms like FalconTST 2.0 rely on large volumes of historical and real-time transaction data to identify patterns and predict future movements in cash flows and exchange rates. Unlike traditional statistical models, which may struggle with irregular or incomplete data, advanced AI systems can differentiate between missing information and true zero activity, reducing the risk of distorted forecasts. This capability is particularly valuable for banks and businesses operating across multiple time zones and markets, where transaction patterns can vary widely due to local holidays, regulatory differences, and sector-specific cycles.
As the use of AI in financial services expands, questions remain about transparency, model validation, and the potential for unintended consequences if forecasts are relied upon without sufficient oversight. While platforms like FalconTST 2.0 offer the promise of greater accuracy and efficiency, their effectiveness depends on the quality of input data, the robustness of underlying algorithms, and the ability of users to interpret and act on model outputs within the broader context of market and regulatory developments.
Time-series forecasting is a core technique in financial risk management, especially for institutions exposed to currency fluctuations. Unlike simple trend analysis, time-series models account for seasonality, irregular events, and structural breaks in data, making them better suited to the realities of international payments and cash management. However, even the most advanced AI models cannot eliminate all uncertainty. Effective risk management requires not only accurate forecasts but also clear governance, scenario planning, and an understanding of the limitations inherent in any predictive system.
