The financial sector is on the brink of a transformation as banks increasingly adopt agentic AI systems, fundamentally changing how they operate. This shift towards AI agents promises not just efficiency but a complete overhaul of legacy systems that currently hinder innovation.
The Rise of Agentic AI
Historically, financial institutions have relied on machine learning and automation, but the emergence of generative AI has sparked a deeper evolution. Unlike traditional tools that produce responses, agentic AI can plan, execute tasks, and achieve specific objectives. This capability positions agentic AI as a key element in the future of banking.
According to the Cambridge Centre for Alternative Finance, 52% of financial firms are already experimenting with or deploying agentic AI. A staggering 81% expect that significant advancements in autonomous agents will occur by 2030. This momentum indicates a major shift in how banks will operate in the near future.
Key Developments in the Banking Sector
Goldman Sachs is collaborating with Anthropic to create AI agents designed to streamline trade processes, transaction accounting, and client onboarding. These areas are essential for banks, which rely heavily on accurate data management and decision-making.
The next generation of banking systems will no longer depend solely on human users navigating static interfaces. Instead, AI agents will act on behalf of various stakeholders — from customers to risk management teams. In this changing landscape, natural language will function as a programming syntax, allowing agents to trigger workflows and manage activities across organizations.
Challenges of Legacy Systems
A major challenge for financial institutions is their reliance on outdated technology built on decades-old code and manual processes. Research by AWS shows that over 40% of business rules may be coded without supporting documentation, leading to what is known as 'agent-blind technical debt'. AI agents require systems that are accessible and interpretable, making it essential to overhaul these legacy systems.
The demand for a comprehensive financial operating system has never been greater. Institutions must move from human-operated, deterministic logic to a model-driven approach that facilitates natural language interaction and agentic operation.
Signs of Progress
Some major banks are beginning to recognize the need for modernization. Citi, for example, is utilizing AI to automate coding and migrate data from legacy systems. According to their CTO, David Griffiths, adopting an agentic approach can accelerate specific tasks by a factor of two to twenty. Similarly, Morgan Stanley has used its DevGen.AI tool to analyze nine million lines of legacy code, converting them into easily understood specifications.
Designing for the Future
As banks embrace AI agents, they must adopt new design principles. Secure pathways for agent access, machine-readable data annotations, and action-oriented capabilities are crucial. Workflows should enhance human collaboration, while business logic must be simplified into policies that agents can interpret. Integrating governance and oversight into autonomous processes will ensure operational integrity.
The transition towards agent-ready systems offers numerous advantages, including increased efficiency and the ability to unlock the full potential of AI in finance. As banks embark on this 'Great Refactor', the emphasis will shift from merely accessing AI models to ensuring that their data and IT infrastructures are ready for the next generation of digital finance. The race to modernize is underway, and those that adapt quickly will likely lead the way in a rapidly evolving sector.
The stories that move AI & crypto markets — before the market reacts.
Free. 7am ET. Five stories. 62,400 readers.
