GenAI in Finance: From Hype to Impact
In the financial sector everyone is already speaking fluent artificial intelligence. But, while leading financial institutions are showing proven business results, many continue to get stuck in pilots and proofs of concept. What is delaying broad adoption of AI, and how can modernization change the picture?
AI solutions have become mainstream in the financial sector. The range of applications is wide and expanding: from risk management, through fraud detection and improving customer experience, to resource management and operational process efficiency. Organizations and financial bodies are no longer satisfied with pilots and technological ‘vision’; they expect measurable business benefit. But are they truly ready for broad implementation that will change the core of operations and lead to long‑term competitive advantage? Not necessarily.
BCG Survey: Only 25% of Financial Institutions Have Incorporated AI Capabilities into Their Strategic Plan
“To enjoy all the advantages AI can offer, and realize the huge business potential inherent in the technology, it must be implemented thoughtfully, preferably across the entire organization,” says Idan Bar, Head of the AI Center of Excellence at Matrix and AI CTO at Matrix Defense. “But many organizations have yet to take this step. According to a BCG survey published this year, only 25% of financial institutions have integrated AI capabilities into their strategic plans. The remaining 75% are still at the stages of stand-alone pilots and proofs of concept, risking loss of relevance as tech-driven competitors continue to lead.”
Productivity, Savings, and Innovation: Leaders are Already Enjoying Results
Organizations that are accelerating adoption of AI are already reaping the rewards. A Bain & Company survey of 109 financial institutions based in the US found that the use of generative AI increased productivity by around 20% in areas such as software development, customer service, and operations, indicating a clear shift from potential to realization. The World Economic Forum reinforces this trend, pointing to operational savings, accelerated processes, and improved customer experience as substantial benefits that are already materializing on the ground.
Examples from the Field: How Financial Organizations are Doing It in Practice
There are also field examples of specific application successes. In fraud detection, for example, Mastercard reported integrating GenAI technology which doubled the rate at which compromised cards were identified, reduced false‑positive alerts by 50%, and tripled the speed of identifying at‑risk merchants.
In credit risk management, more banks are adopting AI‑based tools to uncover complex patterns that cannot be detected with traditional tools or human observation. BNP Paribas reports that as of 2024, it has more than 800 AI‑based use cases in production, across the organization. These include smart risk ratings, automated collection processes, and even virtual agents helping customers handle credit issues. According to the Bank’s report, these projects are expected to generate annual value of around €750 million by 2026, mainly thanks to process efficiency, improved decision‑making, and reduced operational and financial risk.
BNP Paribas is not alone. Banks are no longer satisfied with using AI here and there. For instance, Goldman Sachs has integrated its GS AI Assistant – an AI‑based assistant that supports employees in tasks such as summarizing documents, drafting content, and data analysis. This is a wide‑scale implementation, not a one‑off pilot.
Advancing to the Next Level? Expect Bigger Challenges
“I estimate this trend will only accelerate,” says Idan Bar, “More and more financial organizations will make the shift from proofs of concept to fully-deployed projects. Already today, more banks are incorporating agentic AI – applications that don’t merely meet one-off needs, but rather execute entire task sequences: from receiving a document, through analysis and context understanding, to approval or action execution, while maintaining documentation, compliance, and control.”
But as usage expands, challenges become more complex. “Implementing AI in the financial sector presents unique challenges,” continues Idan Bar. “Regulation, for example, is one of the major ones. The Bank of Israel is a strong regulator that frequently updates its guidelines, and AI systems must be built to adapt to the pace of change and evolving standards. This means designing dynamic and flexible systems from the outset. Alongside this are challenges such as privacy protection, analysis of complex data that combines free text with tabular and numeric information, and dealing with the organizational tech debt that delays many institutions from fully leveraging AI’s potential.”
The Path to Realizing Potential: Infrastructure Modernization
In order to harness AI capabilities to take that business leap, the infrastructure itself must first be addressed and undergo a thorough modernization process. “Everyone talks about innovation, dreams big, and rushes ahead. But once planning and implementation begin, the challenges surface,” says Oren Yosef, SVP of Integration Solutions, Modernization and Financial Digital at Matrix.
“Every organization, especially financial institutions, typically has a foundation of computing systems that has served it for years. These systems contain vast amounts of process knowledge, calculations, and business logic. While they maintain the core of an organization, they are also based on outdated technologies. When an organization decides to integrate AI capabilities – whether for a one-off initiative or wide‑scale implementation – it must rely on the business knowledge already embedded in existing systems. But during project planning, many organizations discover that legacy systems not only make integration difficult, but sometimes even block the possibility of implementing innovation, becoming a business and technological bottleneck.
“Therefore, for many organizations in the financial sector, modernization is not a luxury but a prerequisite for full digital transformation in the AI era. There are a variety of approaches to modernization: rebuilding core systems (Rebuild), replacing them with modern off‑the‑shelf solutions (Replace), converting legacy code to modern code (Refactor), or migrating to cloud environments (Replatform). Choosing the appropriate approach depends on parameters such as system complexity, the existing volume of documentation, change risks, and ROI.”
GenAI Also Accelerates Modernization
According to Oren Yosef: “Whereas in the past modernization projects were considered to be long, complicated processes requiring heavy resources, today AI revolutionizes this domain as well. The introduction of AI‑based tools directly into the modernization process itself has led to significant acceleration. These tools help identify relationships and dependencies, analyze code, generate automatic documentation, and suggest machine‑learning‑based alternatives. Gartner estimates that by 2027, GenAI tools will reduce modernization costs by around 70%! Additionally, there are now modern rule engines that enable decomposition and reconstruction of programming languages and near‑universal migration from any language to any other. Combining these tools with existing AI capabilities helps us to shorten timelines, reduce costs, and achieve higher‑quality outcomes than were possible in the past. In other words, the path to realizing AI’s immense potential begins with using the exact same technology as the engine for accelerating modernization itself.”