The financial services industry is developing its infrastructure along with identifying use cases to deploy agentic AI within its operations. 

The agentic AI market is expected to hit $44.5 billion by 2030, up from $7.3 billion in 2025, a 44% compound annual growth rate, according to the “Beyond Prompts: Building Business for the Age of Agentic AI” report published Oct. 2 by consultancy firm Arthur D. Little.

“This growth reflects a paradigm shift from AI as support to AI as a collaborator, redefining enterprise strategy and business models,” Sri Rajagopal, principal at Arthur D. Little, told FinAi News. 

Agentic AI adoption is still in its early stages, but technology and financial services are leading in experimentation and early deployment among other industries, Rajagopal said. Industries that are data-rich with time sensitive operations will be the biggest winners in deploying agentic AI. 

Some of the most common use cases for agentic AI include: 

• Customer support: AI agents managing multi-turn client interactions, detecting sentiment and escalating intelligently; 

• Risk management and compliance: Real-time monitoring of transactions and adaptive detection of anomalies or regulatory breaches; 

• Portfolio optimization: Allocation of assets using multi-agent systems that learn from real-time market signals; and 

• Operations: Autonomous process orchestration across data systems and APIs to streamline middle- and back-office operations. 

FIs prefer agentic AI over LLMs 

FIs prefer agentic AI over general LLMs due to its autonomy and adaptability, Rajagopal said.  

“Traditional LLMs require constant human prompting and operate within narrow task boundaries,” Rajagopal said. “Agentic AI systems [use LLMs] to execute multistep decisions independently, integrate with external APIs (e.g., trading, CRM, compliance tools) and adapt to evolving inputs.” 

Some FIs are already using agentic AI: 

• OpenAI is working with Stripe for agentic commerce; 

• Wells Fargo is working with Google Cloud to provide insights to clients; 

• Nasdaq, for its AML processes; and 

• JPMorgan, to streamline embedded payments. 

For FIs, this translates to reduced turnaround times, improved compliance accuracy and better complex decision making- effectively moving from prompt-response to goal-driven automation, he said. 

Deploying the tech 

Agentic AI models can be deployed two ways: 

• Horizontal agents, or generalists, are ideal for cross-functional automation including document processing, research, or client support offering scalability and flexibility across departments. 

• Vertical agents, or specialists, provide deep domain precision for regulated workflows like credit risk, compliance monitoring, and fraud detection. 

The solution financial institutions choose depends on the specific use case and what they aim to optimize — whether for scalability, precision or regulatory control, Rajagopal said. 

Revamping bank infrastructure 

As the financial services industry continues to deploy agentic AI, bank infrastructure must be revamped, Rajagopal said. “While technology readiness is improving, significant investment is still required to operationalize agentic systems at scale.” 

Banks must invest in data preparation, cloud modernization, specialized AI talent and continuous maintenance to reap the benefits of agentic AI, Rajagopal said. 

JPMorgan has reported that it aims to invest $18 billion in tech and AI-related initiatives this year, while Bank of America has earmarked $13 billion for its tech and AI initiatives. 

Check out our exclusive new bank industry data here.