LHV Bank has moved from testing artificial intelligence in customer support to using agentic AI in live email workflows, marking another step in the financial sector's shift from AI-assisted work toward more autonomous operations. The UK challenger bank is working with Gradient Labs, an AI company focused on financial-services customer support. FinTech Futures reported on September 28, 2026, that LHV had put fully automated customer-support workflows into production following a six-month pilot. Unlike conventional chatbots that mainly generate responses to prompts, agentic AI systems are designed to perform multi-step tasks within defined boundaries. The development is significant because customer-service emails can involve account information, policies, operational procedures and compliance considerations, making financial institutions particularly sensitive to accuracy and control.
What LHV Bank Has Introduced
LHV Bank is using agentic AI to automate parts of its customer-email operations. The deployment follows a six-month pilot with Gradient Labs, according to FinTech Futures.
The system is designed to handle customer-support workflows rather than simply produce generic text. That distinction matters because financial customer service often requires several steps: understanding a customer's request, finding the relevant information, determining which procedure applies and then preparing an appropriate response.
Agentic AI attempts to coordinate those steps within a controlled environment. Instead of requiring a human employee to perform every stage manually, the software can execute parts of the workflow itself.
What Makes Agentic AI Different
Traditional generative AI is generally used as an assistant. A customer-service employee might provide a question to an AI system, receive a suggested response and then decide whether to send it.
An agentic system is designed to take more responsibility for the workflow. It can potentially determine which tools or information sources are required, perform several actions and complete a task according to predefined rules.
That does not mean the system has unlimited authority. Financial institutions need controls around what an AI agent can access, what decisions it can make and when a human employee must intervene.
The distinction is particularly important in banking because an apparently simple customer request can involve sensitive information or regulated processes.
Why Customer Email Is An Important Use Case
Email remains a major channel for financial-services support. Banks can receive large numbers of repetitive questions involving account administration, product information, payments, documentation and other routine issues.
Many of these interactions follow recognizable procedures. That makes them candidates for automation, provided the institution can maintain appropriate controls and accuracy.
AI can potentially reduce the amount of manual work required for repetitive cases while allowing employees to focus on more complicated customer problems.
However, automation also increases the importance of monitoring. A human employee who makes an error may affect one customer interaction, while an automated system can potentially repeat an error across many interactions.
The Six-Month Pilot
LHV's move into production followed a six-month pilot with Gradient Labs. Pilot periods are particularly important for financial AI systems because they allow banks to evaluate how software behaves against real operational scenarios before expanding its role.
A pilot can reveal issues that are difficult to identify in laboratory testing. Customers may phrase the same question in many different ways, provide incomplete information or combine several requests into one message.
Financial institutions also need to understand how an AI system behaves when it does not have enough information to answer a question.
In those situations, a well-designed system should recognize uncertainty and route the interaction appropriately rather than confidently generating an unsupported answer.
AI And The Changing Role Of Bank Employees
The introduction of agentic AI does not necessarily mean that customer-service teams disappear. Instead, the role of employees can shift toward handling exceptions, complex cases and interactions that require judgment.
Employees may also become responsible for supervising automated workflows, reviewing escalated cases and improving procedures based on recurring problems.
This creates a different operating model from traditional customer service. The objective becomes less about having employees manually process every message and more about combining automated systems with human oversight.
Why Financial Institutions Need Strong Controls
Banking data can include highly sensitive personal and financial information. AI systems operating in this environment therefore need strict access controls and carefully defined permissions.
An agent should only have access to information and actions necessary for its assigned task. The system should also have clear boundaries around actions that could create financial, legal or regulatory consequences.
Auditability is another important requirement. Banks need to understand what an automated system did, what information it used and why an interaction was handled in a particular way.
The Wider Banking Trend
LHV's deployment is part of a wider movement toward agentic AI across financial services. FinTech Futures has reported several other developments involving banks and AI agents, including HSBC's work with agentic orchestration technology and BNP Paribas' expansion of its relationship with Google Cloud around AI capabilities.
The common theme is a shift from using AI purely for analysis or text generation toward embedding AI directly into operational processes.
That shift could eventually affect areas such as compliance, onboarding, fraud investigations, payments operations and internal administration.
Potential Benefits For Banks
- Faster response times: Automated systems can process routine requests without waiting for an employee to become available.
- Consistent workflows: An agent can follow standardized procedures across similar cases.
- Employee productivity: Staff can spend more time on complicated customer issues.
- Scalability: Digital systems can handle increases in customer inquiries without requiring a proportional increase in headcount.
- Operational data: Banks can analyze recurring customer requests and identify areas where products or processes need improvement.
The Risks Of Automation
Agentic AI also introduces risks. A system that misunderstands a customer's request could take the wrong workflow or provide an inaccurate answer.
There are also cybersecurity and privacy considerations. As AI systems gain access to more internal tools and information, protecting those systems becomes increasingly important.
Another challenge is escalation. A bank needs reliable mechanisms for recognizing situations that should be transferred to a human employee.
These challenges explain why financial institutions have generally approached autonomous AI more cautiously than many consumer technology companies.
What To Watch Next
The important question for LHV will be whether the production deployment can expand beyond its initial workflows while maintaining customer satisfaction, accuracy and appropriate controls.
If the system performs reliably, similar approaches could become more common among banks looking to automate customer operations.
The broader significance is that AI is moving deeper into banking infrastructure. The industry is increasingly testing not just whether AI can write a response, but whether an AI system can safely manage an entire business process from beginning to end.
Frequently Asked Questions
What is agentic AI?
Agentic AI refers to systems designed to perform multi-step tasks and take actions within defined rules rather than only generating individual responses.
What is LHV using agentic AI for?
LHV Bank is using agentic AI in customer-email workflows, according to FinTech Futures.
How long was the LHV pilot?
The production deployment followed a six-month pilot with Gradient Labs.
Does agentic AI replace all customer-service employees?
No. Banks can use human employees for complex cases, supervision, exceptions and situations requiring judgment.
Why is AI adoption in banking complicated?
Banking involves sensitive information, regulated activities and financial consequences, so AI systems require strong security, access controls and oversight.
Could other banks adopt similar systems?
Potentially. Several financial institutions are already exploring agentic AI for customer support, compliance and other operational workflows.
Comments
Post a Comment