Financial services has used AI for years. Fraud detection, credit scoring, risk modelling, trading signals, customer segmentation and document processing have all relied on machine learning and automation long before generative AI became mainstream.* But the current wave is different*.

Generative and agentic AI are not just improving isolated tasks. They are changing how financial institutions organize work, serve customers, manage risk, design products and compete for trust.

The industry is moving from digital transformation to intelligence transformation. And that shift requires more than experimentation.

AI Is Becoming a Structural Shift, Not a Side Tool

AI is no longer limited to specialist teams or narrow analytics use cases. Natural language interfaces make advanced capabilities accessible to far more employees and customers. Generative models reduce the cost of producing content, analysis and software. Agentic systems can coordinate multi-step work across tools, teams and systems.

For financial services, this matters because the industry is built on data, language, trust, decisions and process-heavy operations. AI can reshape all of them.

It can help customers receive more personalized advice. It can support faster underwriting, smarter claims handling, better fraud detection, more responsive service and richer financial planning. It can also help institutions operate with more speed and intelligence across the front, middle and back office.

But the same capabilities that create opportunity also introduce risk. When AI starts influencing regulated decisions, customer relationships, liquidity management, fraud controls or advice, institutions need clarity on governance, accountability, data quality, security and human oversight.

Customer Appetite Is Real, but Trust Is Conditional

Customers are increasingly open to AI-enabled financial experiences. One of the strongest signals: 71% of banking consumers globally say they would welcome an AI assistant in their primary bank’s mobile app. But trust has limits: 82% want to approve an agent’s actions before they are executed.

That gap tells a bigger story.

People may want AI-powered convenience, but they still want control. They may welcome proactive guidance, but they do not want invisible automation making consequential financial decisions on their behalf. The future of AI in finance will not be won by removing humans from the loop everywhere. It will be won by designing the right loop.

Financial Firms Are Spending More, but the Question Is Where

AI investment is rising quickly. IDC estimates the global AI market will grow to more than $631 billion by 2028, and recent financial-services-focused research found that 83% of financial services AI practitioners and executives expect AI spending to increase in 2026, with 44% expecting spending to rise by more than 10%.

But more spending does not automatically create more value.

The priority is shifting from buying AI tools to building AI foundations:

• responsible AI guardrails
• enterprise-grade data platforms
• model and agent governance
• AI infrastructure
• workforce upskilling
• leadership capability
• scalable operating models

Financial institutions are learning that pilots are relatively easy. Scaling responsibly is harder.

The Journey Moves Through Three Stages

AI adoption in financial services is moving through three broad stages: piloting, scaling and disruption.

In the pilot stage, firms ask where to start. They experiment with simple processes, internal productivity, copilots and basic AI governance.

In the scaling stage, the questions change. Leaders ask how to accelerate adoption, manage risk, avoid vendor lock-in, redesign workflows and bring AI into customer-facing experiences.

In the disruption stage, AI affects the business model itself. Financial institutions must ask who owns the customer relationship, what products should exist, how services should be priced and how human and digital workers should operate together.

The Focus Is Moving Toward Generative and Agentic AI

Financial institutions are broadening their AI priorities. Data analytics remains important, but generative AI and agentic AI are moving quickly into the center of strategy.

Across the sector, reported focus areas show strong attention on data analytics, generative AI and LLMs, predictive analytics and data processing. Agentic AI is also rising as firms explore autonomous and semi-autonomous workflows.

Different segments show different priorities. Fintech firms, consumer finance companies and capital markets players all have distinct needs, but the direction is clear: AI is moving closer to the core of how financial work happens.

The Real Foundation Is an Enterprise Intelligence Platform

As firms move beyond pilots, they need more than disconnected AI tools. They need an enterprise intelligence platform that connects data, models, workflows, guardrails, auditability, orchestration and human intervention.

This platform matters because financial institutions cannot scale AI safely if every team builds separately.

Scaling requires shared models, reusable data products, access controls, audit logs, orchestration layers, security monitoring and a clear point where humans can intervene. Without that foundation, AI becomes fragmented, with it, AI becomes institutional capability.

The Workforce Must Become Hybrid

AI will not simply replace work in financial services. It will change the design of work. In banking, one estimate suggests 73% of employees’ time could be affected by generative AI, with 39% through automation and 34% through augmentation.

This points to a hybrid workforce where people and AI agents work together.

Humans will increasingly focus on judgment, oversight, relationship-building, exception handling, strategy and accountability. Digital agents will handle analysis, routing, synthesis, monitoring, execution and repetitive coordination, but this does not happen automatically.

Firms need:

• redesigned roles
• skills-based workforce planning
• AI literacy for all employees
• specialist AI capability where needed
• safe experimentation environments
• new performance measures
• leaders who understand both technology and work design

The human side is not secondary. It is the adoption engine.

Agentic AI Is the Next Operating Shift

Agentic AI is different from a chatbot or a simple automation script. It can plan, reason, act, use tools, coordinate with other agents, escalate exceptions and learn from feedback. In financial services, that makes it useful for complex, multi-step workflows.

Potential applications span the full value chain:

• customer onboarding
• KYC and compliance validation
• conversational banking
• personalized offers
• claims processing
• fraud detection
• liquidity forecasting
• credit decisioning
• model risk monitoring
• regulatory impact assessment

The opportunity is large, but so is the need for control.

Agentic systems should not be deployed everywhere simply because they can act. Leaders need to decide where agents should assist, where they can execute under guardrails and where human approval must remain mandatory.

Risk Management Has to Become AI-Native

Financial services cannot treat AI risk as a separate side process. Risk management has to be embedded into the full AI lifecycle: design, build, deploy, operate, monitor, audit and improve.

The most common concerns include data privacy, model reliability, third-party dependencies, operational resilience, cyber threats, legal exposure, reputational risk and financial risk.

One major finding: data privacy and protection ranks as the leading AI-related risk in financial services, cited by 74% of industry players and 80% of regulators.

AI risk management should include:

• clear AI ownership
• model inventory and materiality scoring
• independent validation
• continuous monitoring
• explainability and traceability
• human override paths
• incident response
• third-party assurance
• cyber controls for agent access
• data provenance and quality checks

Responsible AI is no longer just a principle. In financial services, it has to become an operating discipline.

A Two-Speed Model Makes Sense

Financial institutions face two pressures at once. They need near-term value from AI, and they also need long-term transformation.

That requires a two-speed approach.

Speed one delivers visible progress: improved service, faster processing, lower costs and better employee productivity. Speed two builds the foundation: data architecture, intelligence platforms, responsible AI controls, workforce redesign and new business models. The mistake is choosing only one speed.

Short-term wins without structural investment do not scale. Long-term transformation without practical delivery loses momentum.

A Practical Framework for AI Transformation

The strongest approach is not to chase isolated use cases, but to build a staged transformation model.

This framework helps leaders avoid the common pattern of scattered pilots.

The questions become more disciplined:

• What is the AI vision?
• Which domains matter most?
• Where should AI act autonomously?
• Where must human judgment remain central?
• What data and platform capabilities are required?
• How will workflows change?
• How will value be measured?
• How will risk be governed?
• How will the workforce adapt?

Leadership Must Be Shared

AI transformation is not only the job of the CIO, CTO or Chief AI Officer.

It requires shared leadership across the enterprise:

• CEOs and strategy leaders set direction and sponsorship.
• COOs and business leaders redesign processes.
• CFOs measure cost, value and digital capacity.
• CROs and compliance leaders define risk appetite.
• CIOs, CTOs, CDOs, CISOs and CAIOs build the technical backbone.
• CHROs and people leaders redesign roles, skills and workforce models.

AI affects strategy, risk, finance, operations, people and customer trust. The leadership model has to reflect that.

Final Thought

AI in financial services is moving from experimentation to infrastructure. The next phase will not be defined by who launches the most pilots. It will be defined by who can scale AI safely, earn customer trust, redesign work, manage risk and create new value.

The winners will not simply automate old processes. They will build intelligent institutions: organizations where data, models, agents, people, governance and trust work together.

Not AI as a tool, AI as a new operating layer for financial services.

---

Source: World Economic Forum