Banks have moved beyond just talking about AI to handing it real work.
We are moving past the era of the ‘copilot’ that simply assists human agents. We are entering the age of agentic AI, where systems do not just talk, they act inside core processes.
Lloyds Banking Group has just provided the enterprise blueprint for this transition. Generative AI delivered around £50 million of value for the group in 2025. Over 50 AI use cases went live that year, enhancing customer interactions, accelerating query resolution, and boosting frontline colleague support.
Building on that foundation, the bank expects its next wave of agentic AI to generate £100 million in incremental value in 2026 alone. This is not a side project. It is a production-grade shift that changes how the bank serves its 27 million customers. Rohit Dhawan, Director of AI, Lloyds Banking Group confirmed:
“The launch of our new AI-powered Financial Assistant marks a significant step forward in how we serve our customers.”
From ‘Chatting’ To ‘Doing’
Most legacy chatbots are retrieval engines. They find a policy document or an FAQ entry and push it to the customer. Agentic AI is different because it connects to real systems and executes multi-step work under clear guardrails.
Lloyds is launching what it calls the UK’s first agentic financial assistant. It helps customers with routine tasks like card management and subscription blocking. It also supports more active financial coaching, drawing on live transaction data and personalized rules.
That matters because customers do not wake up wanting a better chatbot. They want a faster mortgage change, a dispute resolved, or a clear view of their money. By letting AI perform actions, Lloyds is stripping out the administrative overhead that used to sit with customers and contact center teams.
The operational changes behind those value headlines are concrete. Lloyds reports that income verification for mortgages has shifted from a process that took days to one that takes seconds. Complex fraud and dispute workflows that once stretched across weeks are now resolved in days.
Inside the bank, its Athena assistant supports around 40,000 colleagues. Athena cuts the time spent searching for policies and information by roughly 66 percent. Colleagues can spend more time actually solving problems and less time hunting for the rulebook.
This is the core promise of the hybrid workforce. AI handles the heavy lifting of retrieval, checks, and straight-through processing, while humans focus on judgment, empathy, and edge cases.
The Human Element: Reskilling 67,000 People
When AI agents take on the simpler interactions, the work that reaches a human becomes more complex. Call mix shifts. Cognitive load rises. If leaders do not plan for that shift, burnout risks increase.
Lloyds is responding with a large-scale skills program. Its new AI Academy is designed to reskill all 67,000 colleagues, from the frontline to the executive team. Ron van Kemenade, Group Chief Operating Officer, Lloyds Banking Group, confirmed:
“We are committed to equipping our colleagues with the skills they need to thrive in a digital-first world.”
Group Chief Executive Charlie Nunn has been clear that this is not optional. In recent comments, he warned that bankers will need to “reskill themselves” if they want to thrive in an AI-driven market. His message travels well beyond banking. Any CX organisation that introduces agentic AI without upgrading skills and roles is creating a gap that humans will feel.
For CX leaders, this is the pattern to watch. You introduce agents that can act. You redesign work so humans supervise, coach, and handle nuance. Then you give those humans enough training and support to succeed in the new role.
A Wider Shift Across Financial Services
Lloyds is not an outlier. A broader shift is underway as financial services firms test how far they can safely push AI into action.
Mastercard: Readying Enterprises For The ‘Agentic Era’
At a global level, Mastercard has launched Agent Suite, a set of services built to help enterprises design, test, and deploy agentic AI at scale.
According to the company, a third of enterprise software applications could incorporate agentic AI by 2028. Mastercard expects a ‘significant percentage’ of customer interactions and operational tasks to be supported by AI agents by 2030. Agent Suite is positioned as a readiness layer so banks and merchants can move faster without losing control.
Use cases range from intelligent product discovery for banks to conversational shopping agents for merchants. In both cases, agents do not just answer questions. They surface offers, explain trade-offs, and trigger downstream actions, all inside strong privacy and responsible AI guardrails.
For CX and contact center teams, this is a signal. Payments and commerce journeys will increasingly involve AI actors that can make contextual decisions in real time, not just route calls.
Bank Of Valletta And Oracle: An ‘AI-First’ Operating Model
On a different scale, Bank of Valletta is working with Oracle Financial Services to move toward what it calls an AI-first operating model.
By mid-2026, the $19 billion Malta-based bank plans to deploy Oracle AI agents across retail, corporate, payments, and insurance. These agents are set to streamline loan originations, credit decisions, and customer engagement, and to reinforce compliance and risk management. According to Sovan Shatpathy, SVP / CPO, Oracle Financial Services:
“By leveraging Oracle’s AI agents and architecture, customers can transform labor-intensive workflows into automated, straight-through processes, delegating the heavy lifting of tasks such as data synthesis, summarization and contract creation to AI.”
Crucially, Oracle’s agents are designed with explicit human-in-the-loop checkpoints and regulatory guardrails. Every action is checked against codified rules, not opaque logic. That is the same design pattern we see at Lloyds and Mastercard: AI does the work, humans own the risk.
CommBank: Governance As The Other Half Of Agentic
In Australia, Commonwealth Bank has formalized an enterprise-wide AI governance framework as adoption accelerates across the organisation.
The bank recently set out how it ideates, develops, deploys, and manages AI, with a focus on responsible practices. It uses AI to reduce scams and fraud, strengthen cyber security, and enhance experiences across more than 20 million daily payments, while sending over 40,000 proactive fraud alerts to customers each day.
CommBank’s report details six AI principles, including fairness, transparency, privacy, and accountability. It also highlights tools and guardrails that teams must use when designing and running AI systems.
A key takeaway for CX leaders is that agentic AI cannot sit outside governance. If an AI assistant is allowed to change an account, update an address, or approve a loan, then risk, compliance, and customer teams all need shared visibility of how that agent is built and monitored.
What This Means For CX And Contact Center Leaders
For CX and contact center teams, these banking examples are not abstract. They point toward a new operating model that links front-office journeys with powerful AI in the middle and back office.
We can already see a similar pattern emerging in mainstream CX technology. One recent example comes from Nationwide Building Society, which selected AWS and Amazon Connect to personalize service for 17 million members.
As CX Today reported, Nationwide is using Amazon Connect as a cloud contact center foundation and is layering AI-driven capabilities on top, including a Call Checker service designed to protect customers from impersonation scams. That is a live example of AI acting inside the customer journey, not simply sitting as a passive recommendation engine.
You can read that coverage here: Nationwide Selects AWS And Amazon Connect To Personalise CX For 17 Million Members.
Across these stories, a few patterns stand out for CX leaders:
- AI agents are moving closer to the core. They sit inside lending, fraud, payments, and contact centers, not just on a website.
- Human roles are evolving. Agents become supervisors, problem solvers, and exception handlers rather than script followers.
- Governance is non‑negotiable. CommBank, Mastercard, and Lloyds all show that trust, safety, and explainability must grow with capability.
Trust In The Machine, Confidence In The Team
The biggest barrier to agentic AI in financial services is rarely the model. It is trust. Customers need to know that an AI-driven action is safe. Colleagues need to trust that the system will not undermine them. Regulators need to trust that controls are more than a slide in a deck.
Lloyds handles this by keeping humans in the loop for complex decisions. Bank of Valletta and Oracle embed regulatory rules and human checkpoints into the agent architecture. CommBank publishes its principles and governance model so stakeholders can see how AI is used.
In each case, the AI handles the data-heavy, rules-driven work. Humans step in when empathy, negotiation, or broad judgment is required. That balance is where the hybrid workforce becomes both productive and humane.
The Future Of Hybrid CX Teams
The Lloyds case study, backed by moves from Mastercard, Bank of Valletta, CommBank, and others, shows that the hybrid workforce is no longer a theory. It is here.
Organisations that succeed will treat AI agents as part of the team. They will design clear swimlanes, invest in skills, and connect AI actions to real customer outcomes. They will also resist the urge to automate for its own sake, and instead focus on moments where speed, accuracy, and care all matter at once.
If you are responsible for CX or contact center strategy, the question is not whether agentic AI will arrive in your operation. It is whether you and your people will be ready when it does.
Join the conversation: Join our LinkedIn community (40,000+ members): https://www.linkedin.com/groups/1951190/
Get the weekly rundown: Subscribe to our newsletter: http://cxtoday.com/sign-up
