LHV Bank is testing whether agentic artificial intelligence can help its retail customer service teams handle email enquiries more efficiently, under a proof of concept with Gradient Labs.
The London-based bank said the trial will examine how AI systems that can plan, reason and take controlled actions across set workflows might support customer support teams, particularly in email-based communications.
The proof of concept will focus on whether the technology can improve response quality and turnaround times, while remaining subject to human oversight and the bank’s governance, risk and regulatory requirements.
Gradient Labs, which provides AI support tools for financial services firms, will work with LHV Bank to test selected customer service use cases in what the companies described as a controlled environment.
LHV Bank said the work would place emphasis on explainability, auditability and accountability as it considers how advanced AI systems could be used in customer support.
Kris Brewster (pictured), interim chief executive of LHV Bank, said: “Customer support is a critical touchpoint for retail banking, and the potential use of agentic AI in this area is an exciting opportunity to enhance our service offering.
“This proof of concept allows us to explore how agentic AI could support our teams with email-based customer enquiries, while keeping human judgement and customer protection firmly in place.”
Dimitri Masin, chief executive at Gradient Labs, added: “We’re pleased to collaborate with LHV Bank on this initiative.
“With retail banking, we know how important it is to deliver speed, accuracy, and empathy in one-on-one customer interactions.
“Being able to do it at scale, and safely within this highly regulated environment, is a huge leap forward for customer support in financial services.”
LHV Bank said the project is exploratory and will not involve fully automated decision-making for retail banking customers.
The bank added that any findings from the trial would help shape its future approach to AI-enabled customer support, as well as internal frameworks for advanced AI systems.
