AI’s impact on banking, part 3: How banks can future proof their infrastructure

In today’s volatile world, resilience has become more than just a buzzword for banks—it’s a necessity. As shifting political, fiscal, and regulatory landscapes create uncertainty, the ability to adapt, protect, and recover from disruption is now a defining feature of long-term success.

According to recent CGI 2025 Voice of Our Clients (VOC) data, the changing political landscape has created uncertainty, jittery markets, and more challenging decision-making. Against this backdrop of global unrest and tightening regulation, particularly with frameworks like DORA (Digital Operational Resilience Act) in Europe, financial institutions are doubling down on operational resilience.

However, traditional approaches can only take us so far. To truly future-proof operations, banks must embrace AI—not only as a tool for transformation but as a guardian of resilience itself.

The growing imperative of resilient banking

Based on 2025 VOC results, managing risk and compliance—a key to resiliency—now ranks as the #2 business priority across all of banking and strengthening cybersecurity, regulatory compliance, and risk management ranks as the #4 IT priority.

Banks are already using AI to streamline onboarding, personalize services, and even generate code. So, why not extend that capability to protecting critical infrastructure?

AI offers unique advantages: it can identify risks faster, analyze vast datasets in real time, and predict potential disruptions before they escalate. For banks navigating increasingly complex ecosystems, this shift from reactive recovery to proactive prevention is transformational.

AI’s two key roles in building resilient banks

There are two main areas where AI can play a pivotal role in advancing resilience:

1. Predicting outages before they happen

AI systems can continuously monitor infrastructure to detect anomalies in real time, whether through system logs, application performance data, or network traffic patterns. Machine learning models can learn what “normal” looks like and quickly flag deviations that might indicate impending failures or cyberattacks.

Beyond detection, banks can employ AI-powered predictive maintenance, like what’s used in manufacturing. By analyzing hardware and software performance over time, these models can anticipate failing servers, overloaded APIs, or other weak points, helping IT teams address issues before they cause downtime.

AI can even simulate stress scenarios—like surges in transaction volumes—using digital twins of systems to test resilience in a safe, controlled environment.

2. Evaluating the strength of ecosystem partners

Under DORA and similar frameworks, resilience extends beyond a bank’s internal operations. Financial institutions depend on vast partner ecosystems—cloud providers, fintechs, and other third parties—that all contribute to operational stability.

AI can be used to evaluate partner infrastructure by developing automated risk scoring models that assess cybersecurity posture, uptime, and compliance status. This not only highlights vulnerabilities but provides a data-driven way to manage third-party risk.

Another powerful approach is AI-based network graph analysis, which maps the interdependencies between systems and partners. This can help banks identify single points of failure, especially in scenarios like large-scale cloud outages.

Additionally, natural language processing (NLP) can scan regulatory filings, contracts, and even social media for early signs of instability among partners, giving banks a proactive warning system.

Three key recommendations for banks

So, what should banks do next? Here are three actionable recommendations:

1. Start with the essentials.
   Identify the most critical components of your infrastructure and deploy AI agents to monitor anomalies, perform predictive maintenance, and conduct stress testing.
2. Assess your ecosystem.
   Don’t stop at internal systems. Evaluate the resilience of your partners and vendors, especially those in regulated markets like Europe, where DORA allows for more active testing of third-party systems.
3. Plan for the inevitable.
   Even the best systems can fail. Develop Plan B and Plan C contingencies, such as dual-cloud strategies, to ensure seamless failover when outages occur. Redundancy isn’t optional; it’s essential.

AI as the next frontier of operational resilience

AI can—and really should—be used to help you predict and prepare for your infrastructure future. Whether evaluating delivery models, monitoring compliance, or managing third-party risk, AI enables banks to act faster, smarter, and more efficiently than ever before.

But, with great power comes great responsibility. Responsible AI use and human-in-the-loop governance remain non-negotiable. AI may be the future of resilience, but it’s humans who will define how effectively it’s applied.

In the end, AI isn’t just helping banks stay resilient—it’s helping them evolve.

CGI is working with banks to drive personalization through AI. To learn more about our work, feel free to contact me, or visit cgi.com to explore our banking and AI capabilities. Also, read my first two blogs in this series covering onboarding (part 1) and personalization (part 2).