Global investment bank J.P. Morgan plans a global rollout of an AI-based trading platform after a successful test in the European market by the end of the year, The Financial Times reports.

Dubbed LOXM, the platform uses the AI discipline of Deep Reinforcement Learning that draws lessons from billions of historic and simulated trades to execute client orders with maximum speed and at the best price, according to J.P. Morgan executives.

“Such customization was previously implemented by humans, but now the AI machine is able to do it on a much larger and more efficient scale,” David Fellah, head of the EMEA Linear Quant Research Group at J.P. Morgan, told The Financial Times.

J.P. Morgan limited the platform trading behavior while it learned and made sure that it operates within the firm’s general electronic risk framework, which is overseen by internal control groups and validated by regulators, he added.

Bank executive see DRL eventually taking on the jobs of market making and hedging automatically as the technology matures.

On the retail front, the bank could use LOXM to execute client trades once the platform has learned a customer’s trading behavior and reactions only after the customers grant their permission to use LOXM.