Recently, the Postal Savings Bank of China (PSBC) officially launched a trading robot in the currency market. Relying on the platform iDeal of China Foreign Exchange Trade System (CFETS), PSBC applied AI technology and enabled the robot to possess intelligent online inquiry and reply capabilities for pledge-style repo transactions. This innovation has addressed the challenge of pre-check and reserve of lines of credit before trading, and tackled the most complex and time-consuming process for traders in price inquiry, hence the significantly improved efficiency and quality of PSBC’s local currency market transactions and the accelerated implementation of its business transformation strategy in the financial market.

The robot integrates multiple machine learning and natural language processing technologies. It can simulate traders’ inquiry scenarios in the market, conducting smart conversations of human-computer interaction and online inquiries. Once traders set up their daily trading plans, the robot can replace them in conducting intelligent inquiries and responses. By identifying the intention of counterparties and relevant elements of transactions through conversations, it matches the funding needs with the trading strategies and plans in real time, automatically generates inquiry intention forms and sends them to the counterparties. When intentions are reached, it will convert the forms into pending orders for confirmation.

During the intelligent inquiry, the robot can link in real time with the trading system’s risk control module for integrated risk control throughout the whole process from pre-event counterparty access verification, during-event risk control check to post-event inquiry content analysis. Human intervention can be carried out at any time. When inquiry intentions are reached, the robot will automatically match the inquiry orders with formal trade orders, and execute real-time tracing for compliance management. Moreover, it can check credit in real time and synchronize with the credit management module, supporting pre-check and reserve for pledge of deposits business during the inquiry period, while recommending issuers with sufficient lines of credit to counterparties. In the course of inquiry, the robot monitors and queries each inquiry dialogue in real time, sums up the current number, volume and turnover of inquiries and other trade information, and displays the information visually so that traders can check the overall inquiry process easily and intervene in abnormal dialogues.

It is said that this robot can significantly reduce manual workload, and give play to its advantages of parallel processing to conduct differentiated inquiries with several counterparties simultaneously based on custom plans. During the trial operation, the robot handled inquiries from over 300 institutions, with a total inquiry volume of nearly one trillion yuan (about 142.6 billion US dollars), and succeeded in cost reduction and efficiency improvement.