AI Agents Drive Revenue Growth in Financial Services as 81% of Firms Scale Adoption

By Burley Kawasaki, SVP of Industries, Creatio

As AI moves beyond experimentation, financial institutions need to rethink how they serve customers, organise work and grow their business

A financial institution can process an application faster without making the customer’s experience noticeably better. Documents may arrive sooner, yet the application can still wait for review. A relationship manager may receive an AI-generated summary but lack the authority or information needed to act. Individual tasks improve while the wider process remains difficult to navigate.

That gap becomes more consequential as AI spreads across financial services. The next stage of adoption will depend on how effectively institutions use AI to expand their capacity to grow—helping more customers complete onboarding, giving more businesses consistent relationship support and acting on opportunities that stretched teams cannot pursue today.

Creatio’s Financial Services Edition of The State of Enterprise AI Adoption Report suggests that this transition is already underway. Based on responses from 192 financial services IT and business decision-makers, Creatio's research finds that 80% of surveyed organisations have moved beyond AI experimentation. Meanwhile, 81% are already scaling AI agents across departments or expect to within twelve months.

These findings raise a question for financial services leaders: how should the business change when people and AI agents can accomplish more together?

Connecting adoption to growth

The report’s revenue findings offer one indication of what deeper adoption could mean. Among financial institutions automating more than half their workflows with AI, 44% report increased revenue from AI during the previous six months. Among those automating less than 10%, the figure is 13%.

Those figures describe the proportion of respondents reporting a benefit, rather than the size of any revenue increase. They also show an association between adoption depth and revenue outcomes, without establishing that automation alone caused growth.

Nevertheless, the difference gives institutions a reason to examine where AI agents in banking could support their commercial priorities. Faster onboarding might help more customers complete an application. Better preparation could allow relationship managers to spend more time understanding a business customer’s plans. Earlier identification of a service problem could help an institution retain a valuable relationship.

Each possibility requires a clear explanation of how an operational improvement could contribute to growth. That connection should shape the investment from the beginning, including the workflow selected, the people involved and the outcomes used to judge progress.

Institutions also need to distinguish ambition from readiness. The report classifies just 19% of respondents as AI Adoption Leaders, using a score that combines advanced AI capabilities with agent adoption across business functions. That leaves considerable scope for institutions already using AI to develop a more coordinated approach, even after individual deployments have demonstrated value.

Making additional capacity count

Workforce expectations suggest that many institutions are thinking beyond reducing the cost of existing activity. The survey finds that 87% expect AI agents to augment teams, expand capacity or create new roles, while 12% anticipate significant headcount reductions.

These are expectations rather than employment forecasts. Their significance lies in the management decisions they bring into view. When agents take responsibility for routine work—from gathering information to following up with customers and advancing applications within defined permissions—institutions need a deliberate plan for how employees will use the time released.

A credit union could help more members complete applications and navigate complicated financial decisions. A commercial bank could extend proactive relationship support to smaller business customers who previously received limited attention. An operations team could resolve recurring exceptions that delay account openings or funding.

Realising those opportunities requires more than introducing a tool. Roles, objectives and training need to reflect the new division of work. Employees must understand how to assess an agent’s output, recognise uncertainty and intervene when a situation needs human judgement. Managers need to connect that capability to a service or growth objective.

Designing the whole customer journey

The practical challenge is coordinating work across people, agents and existing systems. In financial services, a process often spans several departments, with different responsibilities for customer relationships, operations, technology and risk.

Consider how a CRM with AI agents could support a business lending enquiry. An agent could assemble customer information, request missing documents, update the application record and route the completed case for assessment, following up with the customer as it progresses. Employees would still need to resolve inconsistencies, understand the customer’s circumstances and exercise judgement within the institution’s policies.

The benefit depends on how those activities fit together. The process needs clear handoffs, reliable information and a route for handling exceptions. Otherwise, faster preparation may simply move a queue further downstream, leaving customers waiting and employees correcting avoidable mistakes.

This makes workflow design a shared leadership responsibility. Business teams can identify where delays or weak service affect customers. Technology teams can establish what agents can reliably do. Risk and compliance teams can help define appropriate controls. Employees handling the work can explain where the proposed design needs to accommodate real situations.

Building on a dependable foundation

Creatio's report identifies substantial obstacles to that coordination. When asked about barriers to scaling AI across their organisations, 52% of respondents cite security concerns, 47% point to data quality and system integration, and 43% identify skills and expertise.

These findings place familiar operational issues at the centre of the AI discussion. An agent preparing information for a relationship manager needs access to accurate customer records in systems such as banking CRM software. One supporting a service process needs to distinguish current instructions from outdated material. Teams need to understand which systems hold authoritative information and how changes flow between them.

Institutions can address these requirements through the workflows they choose to expand. A clearly defined process provides a practical setting for improving data access, clarifying ownership and testing integration. The experience can then inform subsequent deployments. Expansion becomes a repeatable organisational capability, supported by lessons from work already in production.

Giving governance an operational role

Governance becomes particularly important as agents move from providing information to taking action. Preparing a customer summary and changing a customer record involve different responsibilities. An institution needs to define the boundaries of each agent’s authority accordingly.

Creatio's survey finds that 34% of financial institutions have fully established AI governance. Among that group, 72% feel very confident about scaling AI, compared with 23% of those still developing a framework.

The relationship does not prove that governance creates confidence. It does suggest that clear controls and the ability to expand AI can develop together. Teams have a stronger basis for proceeding when they understand what an agent may access, which actions require approval and who is accountable for the result.

Those decisions should be part of process design. Permissions, human review, activity records and escalation arrangements need to match the work being performed. Controls also need to evolve as institutions learn from deployment, change their processes or give agents additional responsibilities.

Customer communication deserves attention too. When an agent supports a decision or interaction, employees should be equipped to explain the process appropriately and resolve concerns. Maintaining that ability matters for institutions whose customer relationships depend on confidence in both the quality of service and the people responsible for delivering it.

Measuring what changes for the institution

Efficiency remains a major reason for pursuing AI agents, cited by 57% of surveyed institutions, followed by productivity at 41%. Both are useful starting points, particularly where employees spend considerable time gathering information or completing repetitive administration.

However, a business case needs measures that follow the intended benefit through to its outcome. An onboarding initiative could track completion rates and customer satisfaction alongside processing time. A relationship management initiative using an enterprise CRM could examine engagement and retention alongside the time spent preparing for meetings.

These measures need clear ownership and a baseline established before deployment. Teams can then assess whether an apparent productivity gain improves service, creates useful capacity or contributes to commercial performance. They can also identify where a process needs further work before expansion.

Regular reviews should bring business, technology and risk owners together around that evidence. A promising result may justify broader deployment; disappointing results may require changes to data, controls or workflow design. Both outcomes provide information for the next investment decision across the institution as adoption expands.

For financial institutions, the next phase of AI adoption will be shaped by these choices. Wider deployment creates opportunities to serve customers more effectively and pursue business that previously stretched available resources. Capturing that opportunity requires leaders to connect agents to dependable workflows, give employees clear responsibilities and evaluate results against business goals. As adoption becomes more common, those capabilities will help determine which institutions turn AI investment into sustained growth.