Backbase Conversational Banking Puts AI Agents in Banks

Backbase Conversational Banking has gone live, putting AI agents that can actually execute banking tasks inside the customer and employee experiences of more than 10 banks. The launch marks a shift in engagement banking software: agents that move beyond answering questions and start completing transactions within bank-defined rules.

According to Finovate, the tool runs on Backbase's AI-native Banking OS and allows customers to describe what they need in plain language — blocking and replacing a lost payment card, checking a payment, or progressing a service request — while the Backbase Conversational Banking agent carries out the permitted steps across connected banking systems. When a request requires human judgment or further investigation, employees can pick up where the agent left off, reviewing a summary of the work already completed.

From chatbots to governed execution in Backbase Conversational Banking

The release comes 18 months after Backbase unveiled its AI-powered banking platform aimed at customer engagement through self-service and real-time support. Backbase Conversational Banking takes that a step further by letting agents execute banking actions on behalf of consumers and employees, provided those actions fall within pre-defined permissions, policies, and approvals set by each bank.

The agents behind Backbase Conversational Banking work alongside existing digital banking stacks, so banks do not need to replace their mobile or online platforms to deploy the new capability. Customers use natural language prompts, and the Backbase Conversational Banking agents draw on customer context, permissions, and connected systems to act. According to itbrief.co.uk, each completed action creates a decision token that records what was approved, by whom, and under which policy — giving compliance teams a full audit trail of agent activity.

Governance is deliberately embedded in the execution layer. Banks decide which data, knowledge sources, and tools their agents can access, which actions they may perform, and where human approval is required. According to fintech.global, actions with significant consequences require explicit customer confirmation, while logs of the context, checks, decisions, and results support a complete audit trail.

Real deployments and what the numbers say

The product is already live at more than 10 financial institutions, and the early metrics are striking. One of South Africa's four largest banks has supported 22 million customer interactions through the system while increasing self-service containment from 20% to 70%, according to Finovate. That means nearly three-quarters of those interactions were resolved without being escalated to a human.

The economics argument is equally ambitious. According to McKinsey & Company research cited by Backbase, agentic AI could cut banks' operating costs by 15% to 20% by automating routine information gathering and freeing staff for advisory work, exceptions, and complex cases. As reported by fintech.global, the pitch is that customers get tasks completed directly through conversational interfaces rather than being sent to a call center or branch for routine requests.

Backbase founder and CEO Jouk Pleiter has framed the moment as the arrival of a long-promised transition: digital banking put the bank in customers' pockets, and the next step is customers simply saying what they need done while agents act on it intelligently. According to Finovate, Pleiter described this as a future in which agents do the work while banks stay in control — a shift he says completely changes both the customer experience and the economics of serving them.

Why banks are betting on Backbase Conversational Banking

The Backbase Conversational Banking launch highlights a broader industry shift from AI-powered tools to AI-powered execution. Chatbots have historically been limited to providing information and directing customers to help articles. By handing routine service requests over to agentic AI, banks can automate the steps required to resolve them — turning AI from a customer service tool into a functional layer of the banking experience.

The timing matters for the wider AI agent ecosystem. Banks have spent years testing AI in customer service and internal operations while wrestling with compliance and approval controls. Stuut's order-to-cash agents and Melius' creative ad agents show how the same agentic wave is reshaping finance operations elsewhere in the enterprise. A vendor with a background in regulated financial infrastructure, already working with more than 120 banks across 50 countries including Navy Federal Credit Union, TD Bank, and KeyBank according to itbrief.co.uk, is well positioned to package governance into the agentic process itself — one of the main hurdles keeping agents out of production.

Whether banks use these capabilities primarily to cut costs or to deliver more personalized service remains an open question. Automating routine requests clearly improves efficiency, but the longer-term opportunity may be using agentic AI to offer more proactive, personalized banking. Backbase's broader portfolio, including tools for identifying customer needs and coordinating service delivery across teams, suggests the company is aiming at both: cheaper operations today, a more capable banking relationship tomorrow. For enterprises watching the agentic shift, Backbase Conversational Banking joins moves like Google's Gemini agent for enterprise work in the race to make AI do work, not just answer questions.