High-yield AI in financial services depends on secure, scalable networks

The financial services industry is now supercharged by AI. Consider the case of a major bank that uses AI-powered chatbots to understand and respond to customer queries, in natural language, with an increasing degree of autonomy. Customer wait times are shorter than ever and overall customer satisfaction is rising.

Beyond the contact center, AI is being used in risk management and compliance, portfolio management, fraud detection, credit scoring and regulatory reporting — it seems the applications are endless.

In a recent IDC webinar with NTT DATA and Cisco, Accelerating AI in FSI with network modernization, we explored how AI is reshaping financial services — and why FSIs must urgently define their strategic stance on it.

Importantly, we considered what it means to think like an AI-first organization. AI isn’t a side project anymore. It’s your next teammate. The shift is clear: your business doesn’t need an AI roadmap — your business roadmap needs AI. From basic automation to agentic, domain-specific collaborators, AI is now a continuum. But unlocking its full potential requires trust, understanding and governance.

Let’s take a closer look at these and other key topics from the webinar.

AI in FSI: Adoption levels, opportunities and risks

AI is the most significant IT trend to have emerged in the past 20 years. Some may argue it’s the biggest we’ve ever seen. Today, most FSIs have adopted AI to some extent. According to an IDC survey*, 50% of banks will adopt GenAI within the next 24 months, whether it’s to improve customer service through hyperpersonalization, boost employee experiences, strengthen their security posture or make operations more efficient.

But, as with most major technological breakthroughs, AI comes with potential risks. Our conversations with FSI clients have uncovered lingering questions and concerns, including:

• Shadow AI (the unsanctioned use of AI tools). Employees want to use these tools to do their jobs faster and more effectively. How do IT teams manage, monitor and govern this usage?
• The exploitation of AI and GenAI by cybercriminals. FSIs are primary targets for cyberattacks. How can they defend themselves against threat actors who are capitalizing on advancing AI technologies to create — and automate — more sophisticated social engineering attacks?
• Putting the appropriate AI governance guardrails in place: How do we use, store and manage our employees’ and customers’ data in a responsible, compliant and ethical way?

FSIs also have to gauge whether their current infrastructure is geared to running AI projects and supporting the associated technologies, workloads and services that underpin them.

To perform at its peak, AI needs a network infrastructure that can deliver high-quality data at speed and scale, not just within the data center but also all the way to edge devices and the cloud — all of which puts significant pressure on infrastructure resilience.

The AI era presents FSIs with an opportunity to reimagine and rearchitect their infrastructure from the ground up.

• ALSO READ → The new frontier: Why agentic AI should be on every CIO’s roadmap

Becoming an AI-first organization: A strategic playbook

NTT DATA and Cisco advocate a six-pillar approach to AI adoption that’s rooted in strategic clarity and operational readiness:

• Strategy and business intent: Start with clear business priorities. Identify where AI can deliver measurable value — for example, through customer insights, fraud detection or operational efficiency.
• Infrastructure readiness and process integration: Assess whether current systems can support AI workloads. Invest in managed services and platforms with built-in AI capabilities and the ability to scale effectively.
• Process integration: AI doesn’t thrive in silos. For it to become a core operating principle, it must be embedded into the workflows, decision-making cycles and operational rhythms of the business.
• Trust and governance: Establish ethical AI frameworks and governance committees (involving legal, HR and security teams) that are founded in zero trust principles and architecture. Ultimately, trust is the currency of business. Build it. Communicate it. Standardize it. Cisco’s Trust Center is one example of how organizations can formalize responsible AI use.
• Data quality: The cornerstone of trustworthy, high-performing AI. Without it, models risk bias, failure and reputational damage. Contextual relevance and traceability keep AI systems resilient, compliant and aligned with stakeholder expectations.
• People and skills: Education is the human core of AI readiness. Looking beyond hiring data scientists or upskilling engineers, it’s also about cultivating an organization-wide mindset shift: from AI as a tool to AI as a collaborator. NTT DATA’s internal “AI academy” is a model for how organizations can upskill their workforce and foster a culture of innovation.

Agentic AI: The next frontier

No discussion about AI would be complete without delving into agentic AI — AI systems that act autonomously within defined parameters.

Cisco is pioneering the Internet of Agents, a framework that allows specialized AI agents from different domains to communicate and collaborate, much like how the Internet Protocol works for the internet. This creates new possibilities for intelligent automation, decentralized decision-making and scalable innovation.

As AI agents take on more repeatable tasks, human roles will evolve toward strategic oversight and creative problem-solving. We emphasized the importance of having humans in the loop — building trust gradually, allowing for human approval in the early stages and increasing automation as confidence in AI systems grows.

• ALSO READ → Smart agents, smarter business: Experience the power of agentic AI

This article includes contributions by Hendrik Blokhuis, Director and CTO for Partner Organization, Europe, Middle East and Africa, at Cisco.

* IDC. AI and GenAI Adoption in Corporate Banking: From Experimentation to Industry Transformation. Document number US52805325. March 2025.