How AI in banking can result in major transformative benefits

Banks have seen significant impacts from GenAI and expect even greater transformation in the next few years.

In brief

• Sixty-one percent of the banking survey respondents already report substantial impacts from their GenAI deployments. 
• Front-office use cases move to the forefront and dominate applications reaching production. 
• Organizational governance and workforce dynamics emerge as key elements of success.

Generative AI (GenAI) continues to rapidly evolve, driving significant investment and optimism across banking institutions. The 2025 edition of the EY-Parthenon Generative AI in Banking survey reveals substantial progress since 2023, with 77% of banks having now actively launched or soft-launched GenAI applications vs. 61% in 2023. Concurrently, interest in agentic AI is gaining traction, attracting resources and attention. 

But for GenAI specifically, 61% of survey respondents already report substantial impacts from their deployments and 89% expect major transformative benefits within the next two years. This highlights a clear shift from proof of concept into production for many organizations and a high degree of optimism for future impact.

The growing confidence in GenAI is further highlighted with 38% of respondents now anticipating full end-to-end automation potential across key functions in the next five years, a substantial increase from 2023.

While enthusiasm for GenAI and the newly emerging agentic AI remains high, most banks still rely on traditional machine learning (ML) and robotic process automation (RPA). Only 28% of automation use cases in development or implemented by banks currently employ GenAI or agentic AI. This shows that banks recognize the need to strategically select the optimal technologies aligned with specific use cases to enhance value — a critical area of strategic planning as they press forward on this journey.

Agentic AI, capable of autonomous decision-making, has quickly entered banking awareness, with 99% familiarity among respondents. However, only 31% have moved toward actual implementation, signaling immense growth potential for first movers in this space.

Banks also reported significant hurdles — data-related challenges, regulatory compliance and execution issues — resulting in high development and implementation failure rates. Addressing these challenges and being mindful of the learnings of the past two years will be crucial to unlocking GenAI’s full strategic value for retail and commercial banking. Nevertheless, significant progress in rolling out GenAI applications has been made in the past two years, with 47% of respondents in 2025 indicating having rolled out GenAI applications compared to 10% in 2023. Additionally, 90% of banks in 2025 are at least in the beta-testing stage or further, compared to just 64% in 2023, indicating that only the slowest adopters have yet to operationalize their GenAI strategy. 

Despite the challenges that come with early investments in new technologies, banks are beginning to reap the rewards. Fifty-eight percent of banks report anticipating a positive revenue uplift of 6% to 20% from GenAI applications. Additionally, 79% of respondents expected further revenue uplifts of 6% to 20% over the next two years. So, while banks are working to improve their use case selection frameworks and reduce failure rates in development and rollout, the investments often pay for themselves with long-term value being realized. 

AI in banking: Embracing centralized governance for enhanced decision-making

Institutions have increasingly embraced structured governance models to navigate the complexities of modern financial landscapes. A significant majority have shifted toward centralized governance approaches, which streamline decision-making and enhance accountability. In contrast, federated governance models, which distribute authority across various business units or regions, have seen a decline in adoption. However, hybrid governance models, which combine elements of both centralized and federated structures, are still utilized by over 40% of banks.

While GenAI has foundational legs across firms, larger banks have also demonstrated greater readiness and capability for adopting agentic AI but have noted significant barriers such as regulatory compliance (71%) and data privacy concerns (67%), necessitating careful planning and robust governance.

Banks will need to continuously apply rigorous prioritization of use cases based on technology fit, value creation potential, data readiness and risk profiles for both GenAI and agentic AI, leveraging broad industry insights into the successes and failures of peer institutions coupled with technologist expertise to enhance success rates.

Key risks, challenges and lessons learned related to AI in banking

Despite robust investment and early successes, some execution challenges remain. Data shows that the development to implementation conversion rate can be low (only 16% of use cases reach full deployment). Failed use cases are common, with 40% of implemented use cases failing to meet expectations. This underscores the need for continued efforts to integrate lessons learned into new initiatives. 

Banks have developed GenAI use cases to achieve benefits, but struggle with early failure rates

Primary barriers to further progress include regulatory compliance challenges (26%), data privacy concerns (22%) and limited high-quality data access (21%). To achieve greater success, banks clearly recognize the necessity for enhanced data governance (cited by 79% of respondents) and earlier, deeper stakeholder engagement (71%) to reduce these barriers.

Recommendations for banks moving into the next horizon

As the banking sector increasingly integrates GenAI into its business models, it is crucial for institutions to adopt a strategic framework that addresses opportunities and challenges. The following recommendations outline essential steps for leveraging these advanced technologies to drive innovation, improve operational efficiency and provide sustainable growth in a competitive landscape.

Banks have the opportunity to rapidly embrace GenAI and explore agentic AI to transform operations, drive innovation and strategically position themselves for future growth. By focusing on enhancing data governance, engaging stakeholders and prioritizing external partnerships, banks can effectively navigate the complexities of implementation. While challenges persist, proactive governance, strategic alignment and careful use case prioritization offer clear pathways to realize GenAI’s full potential.

Nahom Brhane, Senior, Benjellica Leslie-Jones, Senior, Pedro Fernandez, Senior, and Yusuf Azizi, Senior, were contributors to this article.