How operating model misalignment stalls AI in banking and insurance

Financial organizations launch ambitious AI plans and pilots, but progress stalls because their operating models aren’t built to scale AI.

In brief:

• AI initiatives often fail not because of tech limits but because operating models are misaligned with how AI truly works.
• Success requires proper governance, standardized data and a culture built for AI agility.
• Effective alignment enhances the impact of technology, but it is the operating model that empowers success.

Across the financial services industry, the same pattern repeats: ambitious AI plans, impressive proofs of concept and then a stall. Models that work in a lab rarely make it into production, and promising pilots fail to scale.

In a recent report from MIT, it was shown that 95% of organizations are getting zero returns from their AI investments, and only 5% of integrated AI pilots are extracting millions in value, while the vast majority remain stuck with no measurable P&L impact (source: State of AI in Business, MIT, June 2025).

Meanwhile, boards grow frustrated that the return on AI investments remains small. Most banks and insurers are currently investing in and exploring their technical foundations for AI. They have the infrastructure and hire the right talent but struggle to translate the investments into measurable outcomes. The failure is not due to algorithms or cloud platforms but rather a failure of alignment between technology, business processes and the operating model of organizations.

The problem begins once the AI teams operate in isolation. Usually, teams are placed within innovation hubs or IT departments. The result is organizational friction, where business lines that own the P&L lack involvement, risk and compliance functions apply traditional rules to systems because they are designed to manage static models, decision-making remains siloed, and the employees rarely have access to insights that AI could generate.

Why traditional operating models don’t fit

Banks and insurers were designed for reliability, control and risk mitigation. These are the traits that built trust in these organizations. However, this also led to hierarchies, rigid governance structures and slow decision-making processes. AI demands something different and thrives in environments that are iterative, cross-functional and adaptive. Legacy operating models of financial institutions were not built to function this way, and the result is an ongoing tension between AI’s dynamic nature and the static design of organizations.

Becoming AI-ready does not mean that you need to dismantle the entire institution’s foundations. It does however mean that you need to redesign the operating model so that governance, talent and culture enable AI innovation. According to our experience, we have identified five critical traits that successful organizations share:

1. Clear ownership and governance
   AI success begins with accountability. AI “councils” or “model oversight committees” are becoming more common. These bodies do not only approve models but also define ownership of outcomes and establish transparent standards for ethics, explainability and compliance.
   
2. Business-embedded AI teams
   Firms are moving toward federated models, e.g., where AI specialists sit within core business functions (such as lending, claims, underwriting) and collaborate daily with domain experts. Furthermore, more financial institutions promote AI use to their employees and focus on upskilling their teams.
   This closer collaboration between the business and the AI specialists, as well as upskilling of all employees, creates a structure that increases the understanding of how AI can be used and ensures that models solve real problems and are adopted quickly.
3. End-to-end data enablement
   AI cannot thrive on fragmented and inconsistent data. This is why institutions need to treat data as a shared asset and a strategic infrastructure rather than a byproduct. There is a strong need for unified data platforms, standardized data and clear governance that balances accessibility with control for AI to be useful and generate value.
4. Continuous model management
   Deployment is not the finish line; it is the starting point. It is important to have practices that allow models to be monitored, retrained when needed and continuously audited throughout their lifecycle in order to ensure accuracy, fairness and compliance even as conditions change.
5. A culture of experimentation
   One of the hardest shifts for organizations is the cultural shift. AI progress is dependent on experimentation, running pilots, learning, and potentially dismissing quickly and scaling when possible. AI-ready culture means measured risk-taking and data-driven iteration supported by leaders who understand that not every experiment needs to succeed to deliver value. Testing, failing, testing again and succeeding is key. Being afraid to fail needs to be eradicated with the idea of being excited to try.

Ultimately, building operating models that are AI-ready is a technology project but with an organizational transformation at the core. Leaders need to take on a more holistic view and align around strategy, incentives and behaviors around the use of data. That means, for instance: setting KPIs that are tied to AI-driven outcomes and not just adoption rates, ensuring that all executives understand the basics of AI (including both possibilities and limitations), building trust with regulators through transparency in model design and decision logic, and fostering collaboration across data, business and risk functions to safeguard compliance while accelerating innovation.

One area within financial services where AI has the potential to significantly enhance is the fight against financial crime by improving the detection and prevention of fraud and money laundering. As fraudsters become more sophisticated in leveraging AI to their advantage, it is essential for banks and insurers to adopt similar technologies to remain competitive and effectively combat these evolving threats. If you are curious about AI use within financial crime, please read more here.

In conclusion, the potential for AI in financial services is undeniable. Aligning operating models to AI and teaming, rather than AI teams working in isolation, will be the key factor for success. Bankers and insurers that recognize this shift and build their organizations accordingly will see AI move from isolated, failed experiments to enterprise-scale transformations. Are you interested in learning more about how your organization can get the best value out of AI and how to ensure that AI is integrated with the day-to-day business? The EY organization has many professionals who help our clients with similar journeys and AI transformations. Please contact us for more information.