Dara Sosulski, head of AI and model management at HSBC, has said that the bank is entering a mature phase of AI implementation and is now heading towards “real ROI”.

Speaking at FTT Festival in London on Monday, the AI expert said that the bank has several AI use cases already in production, adding that there are many more projects in private testing and proof-of-concept stages.

“We believe we are close to achieving significant and meaningful return on investment from our initiatives,” she told the audience.

Sosulski warned that traditional, slow methods are no longer sustainable in the current era of AI, with organisations urgently seeking to implement and leverage AI technologies “as quickly and effectively as possible”.

She said that her team is currently focusing on developing advanced technology platforms that enable rapid AI implementation.

During a panel discussion focused on the ethical and responsible use of AI in financial services, Sosulski explained how in the past, when developing a model such as a credit response system, the bank would create the model, demonstrate its functionality using standard libraries, obtain stakeholder approval, and then submit it to validators and regulators for approval.

“This entire process could take one to two years,” Sosulski said.

The solution, according to Sosulski, lies in developing technology platforms capable of managing rapid AI implementation, with built-in controls, monitoring, testing capabilities and analytics pipelines, which are currently a key focus for the head of AI and model management and her team at HSBC.

“These platforms should be efficient, streamlined, and focus on tracking changes rather than repeatedly rebuilding processes from scratch,” she said.

During the panel, she also explained the complexity of managing AI systems, particularly as they become more advanced, such as agentic systems.

“Organisations must thoroughly understand their AI’s functionality, feel confident about its operations, and be able to justify this confidence to regulators,” she said.

Commenting on how to balance the need to innovate quickly while providing safeguards for ethical AI, she discussed the importance of understanding AI technology from a regulatory perspective.

“Organisations must comprehensively understand how an AI system operates, feel confident about its functionality, and be able to clearly explain and justify that confidence to regulators and stakeholders,” continued Sosulski.

Speaking at the same event on Monday, Lloyds Bank’s group head of AI said that the organisation is on track to meet its target of delivering £50 million worth of incremental value through 50 genAI use cases this year.

Rohit Dhawan said that the bank’s AI journey has been more focused on innovation rather than fully advanced models in production, with Lloyds tracking the “total number of AI advances.”

---