In 2025, the financial services industry will continue to build out its AI, GenAI, cloud delivery, and process automation capabilities. NelsonHall’s recent survey of financial services executives identified that:

• Operational transformation is highly important to 92% of retail banking companies and 93% of commercial banking companies 
• Current operating models and processes at over half of institutions are not highly adequate to support key business lines, including payments, lending, and wealth management.

Banks will change how they pursue their business objectives in 2025, including in the areas of AI/GenAI, cloud delivery, and process automation.

AI/GenAI: operationalizing Small Language Model projects  

In 2024, every bank wanted to investigate GenAI and appear to have an AI strategy. This resulted in many consulting engagements and POCs. Most POCs do not have a high ROI, and very few banks have operationalized any GenAI POCs. Furthermore, banks have discovered that successfully operationalizing GenAI using LLMs requires vast data, processing power, and energy consumption.  

In 2025, most banks will decide that GenAI technology needs greater maturity before they invest heavily in operationalizing the technology internally. However, early adopter banks will move from testing the technology to operationalizing it in a narrowly defined set of use cases. The use cases will use Small Language Models (SLMs) to address internal operations processes such as:

• Software development, including more efficient generation of code documentation
• Hyperpersonalized content for marketing campaigns, such as generating thousands of unique marketing documents and emails personalized to each intended recipient  
• Employee online training programs customized to each employee
• Email/document extraction and knowledge summarization
• Risk management, where banks train apps on special-purpose data sets, such as specific loan types used by regional/local customers, to identify risk characteristics and initiate early remediation activities.

While banks will operationalize GenAI use cases, AI efforts in 2025 will primarily focus on modernizing the data architecture. To improve the data architecture, banks will seek to:

• Define a data security architecture that is effective under a rapidly changing technology environment
• Decide what applications can be used in each part of a bank’s tech ecosystem (e.g., proprietary apps in the customer-facing environment and third-party apps in the internal facing environment)
• Decide where the data will be hosted and who has access.     

Finally, banks will deploy AI embedded in processes to enhance corporate action resolutions, payment dispute resolutions, and prospect and at-risk client identification and prioritization.

Cloud delivery: standardizing faster delivery & embedding domain IP into hyperscaler solutions

Banks have been accelerating their move to the cloud. Over the past two years, most banks have developed target operating models (TOMs) and have adopted a hybrid, multi-cloud strategy. In 2025, banks will focus their cloud activities on:

• Migrating payments and securities trading to the cloud to improve cross-border transaction processing and declining settlement times (on the path to comprehensive real-time settlement)
• Standardizing and coordinating multi-country cloud delivery to provide a single-brand experience to customers
• Building joint solutions with hyperscalers, including:
  • Cybersecurity and fraud solutions with industry customization
  • Frameworks and solutions to deliver custom synthetic data sets
  • Automation of data migration
  • Automation of hollow-the-core processes. 

Process automation for payments and lending  

The adoption of process automation has been accelerating because:

• Cost pressures have accelerated the automation of manual processes
• The conversion of platforms to microservices has delayed the rewriting of solution modules, which will remain legacy for the time being.

In 2025, the focus on process automation will be on automating processes in lending (primarily origination) and payments (settlement, reconciliation, and reporting). This represents a significant shift from focusing on customer contact to transaction execution. By embedding AI in these automated processes, banks can complete transactions much faster, setting the stage for real-time processing, not just near real-time processing.     

Summary

Banks are maturing their approach to using RPA, cloud, and AI in operations. In 2025, they will operationalize these technologies in tightly defined and rigorously controlled domain-specific use cases. GenAI will be deployed to address domain-intense challenges such as hyper-personalized marketing campaigns, employee training, software development, and document summarization. Cloud activities will automate migration, standardize TOMs, and migrate transaction processes. And finally, payments and lending sub-processes will be automated with embedded AI to address domain-specific requirements.