Banks are increasingly deploying AI agents to automate routine tasks and improve operational efficiency
AI is moving quickly. And banks are acting. Right now, global financial institutions are deploying digital agents to handle high-volume, low-complexity work. This shift isn’t about replacing people. It’s about elevating them. Tasks no one really wants to do, vulnerability assessments, routine code fixes, are being handed off to AI systems with precision and speed. That frees up engineering teams to focus on product development, implement new features, and drive better customer experiences.
You want teams building the future, not buried in maintenance tickets. BNY built its own AI agents from the ground up. These agents tackle infrastructure-level work and report their output to human supervisors. Simple issues are resolved automatically. Complex ones get escalated. This isn’t science fiction, it’s operational now.
Deploying agents like this removes friction from day-to-day operations. It also accelerates how fast change can happen. The foundational tech is solid. Early adopters are already showing better output per engineer, cleaner workflows, and happier teams.
Capgemini puts the potential economic value at $450 billion over the next three years. That’s not a vague promise. That’s real money sitting on the table for firms that move first and scale fast.
Leigh-Ann Russell, Chief Information Officer at BNY, summed it up well: “This frees up our engineers to do new feature releases and other fun stuff.” That’s the goal, move your people upstream and let machines handle the mundane.
Most banks are developing AI agents internally rather than purchasing off-the-shelf solutions
Here’s what’s interesting: most banks aren’t buying generic AI tools. They’re building them. In-house development is the dominant strategy now. That’s not just a control issue, it’s a quality one. Leaders want precision systems tuned to their infrastructure, operational scale, and compliance models. If you’re running a system that handles billions in transactions, you don’t just plug in third-party code and hope for the best.
Custom tools give banks more flexibility during integration phases. They can shift priorities quickly, adapt systems as regulatory environments evolve, and keep data pipelines clean. Off-the-shelf AI sounds fast, but it often requires deep customization anyway. Building from the ground up skips that inefficiency.
This also aligns with how technical leaders think about scalability. Build it right from the outset, and future capabilities, like language localization, deeper predictive analytics, or integration with internal LLMs, can be layered in at speed.
Capgemini’s report found that one-third of financial firms are building AI agents internally. Only 16% are going with commercial alternatives. That tells you where serious players are headed.
BNY’s CIO, Leigh-Ann Russell, confirmed their AI tools are homegrown, even used to automate cybersecurity work. This internal expertise pays dividends as the tech scales. You get tighter feedback loops, better uptime, and less lag between innovation and deployment.
If you want AI to power the core, not just sit on the edge, it needs to be yours. Your code, your controls, your roadmap.
Full-scale implementation of AI agents remains limited despite strong pilot and ideation interest
There’s plenty of ambition when it comes to AI agents. Most financial institutions are exploring the tech. They’re piloting tools, running proof-of-concept tests, seeing what sticks. According to Capgemini, 80% of banks are currently at the ideation or testing stage. That’s a strong signal that intent is high.
But here’s the issue, real deployment at scale is rare. Only 10% of firms have actually rolled out AI agents across their operations. That means the industry is still in early adoption mode. It’s not a technology problem. The capabilities are proven. It’s an execution gap.
Moving beyond test environments requires more than enthusiasm. It takes operational readiness, clean data, modern infrastructure, internal alignment between IT and business units. AI agents can’t scale in silos. They need system-level access, and that means organizational transformation, not just tech adoption.
If you’re in the C-suite, understand this: most of your competitors are still figuring out how to cross that line from pilot to production. Whoever gets there first, with a system that works institution-wide, unlocks a meaningful edge, faster workflows, reduced errors, and more responsive product cycles.
The technology is ready. The question is whether internal systems and leadership structures can keep up.
Skills shortages and regulatory compliance are significant barriers to AI agent adoption
Banks want to scale AI. But they’re hitting predictable roadblocks, talent and regulation. Capgemini found that 92% of banks report a sizable skills gap as a key barrier, almost as high as the 96% citing compliance risk. These aren’t small problems. They’re immediate challenges that can stall progress and kill momentum.
On the skills side, these systems need engineers who understand AI architecture, governance, cloud strategy, and secure deployment. That intersection of competencies is thin right now, especially in traditional finance environments. Upskilling and hiring can’t lag behind the tech.
Then there’s regulation. Financial systems deal with complex legal frameworks, data privacy, model transparency, explainability requirements. AI agents that act autonomously can’t be treated as black boxes. Their decisions and internal logic need to be auditable. This takes work. Not just from engineering teams, but from legal, compliance, and risk management leaders who need to define new guardrails.
You can’t approach this casually. If your team doesn’t have deep regulatory context, your AI tools won’t make it past internal controls, or worse, they’ll trigger external scrutiny. At the same time, high oversight shouldn’t mean paralysis. With the right planning, banks can build compliant infrastructure that supports innovation rather than restricts it.
Scaling AI isn’t just about writing better code. It’s about closing the talent gap and building trust, internally and externally, through systems that are both powerful and transparent. Leadership needs to solve for both.
Industries beyond finance are also investing in AI agents, leveraging combined AI and cloud capabilities
This isn’t just a banking shift. The momentum around AI agents is extending across industries. Companies like PepsiCo, Walmart, and Colgate-Palmolive are already deploying these systems. Operations, logistics, supply chain, every function with repeatable, data-heavy workflows is a candidate for automation. AI agents are being tasked with decision support, task execution, and real-time monitoring across enterprise systems.
The common thread enabling this? Cloud infrastructure. The combination of scalable compute and intelligent systems is what’s powering agentic AI outside of financial services. It gives enterprises rapid deployment cycles and the ability to process massive workloads without bottlenecks. That’s what’s making this shift operational, not theoretical.
For executives, the takeaway is clear. You don’t need to sit back and wait to see how finance does it. If you’re running enterprise technology at scale, the building blocks are there to start now, especially if you’re already deeply invested in cloud transformation.
Khokhar, referenced in the Capgemini report, pointed to this intersection as a key driver. AI by itself isn’t enough. Once it’s layered with a robust cloud backbone, you get the kind of operational visibility, data flow, and modular deployment needed to execute across teams and business lines.
Expect the entire enterprise tech stack to be reshaped over the next few years as AI agent systems begin handling not just background tasks, but core workflows. Industry lines don’t matter much when automation connects directly into platform infrastructure.
Long-term strategy and scalability are essential for extracting maximum value from AI agent deployments
Pilots are useful. But isolated results don’t scale impact. Long-term value from AI agents happens when organizations sharpen their operational model, align their technology roadmap, and prepare to scale deliberately.
Leadership teams need to start by defining what success actually looks like. Is the goal operational cost savings? Faster product delivery cycles? Error reduction across technical workflows? Without that clarity, you won’t get sustained ROI because the deployment objectives aren’t tethered to real business impact.
Scalability is not just about adding more AI agents. It’s about shaping the systems around them, governance, monitoring, training, and adjustment. Teams need feedback built into the loop and protocols to manage AI activity at scale. Otherwise, you’ll deploy but never optimize.
High implementation costs tend to invite hesitation, but the time-to-impact can be short if the systems are aligned. The tech is capable. The question is whether the rollout model is right.
Khokhar emphasized the importance of long-term strategic planning. Leaders need to be thinking about what their operating model looks like once AI agents are fully integrated, not just what the next pilot will test. That mindset shift is what creates compounding benefits over time.
Organizations that take AI seriously at the structural level will see the gains. Not just early, but consistently. That’s the difference between experimentation and transformation.
Key takeaways for leaders
- AI agents free up skilled talent: Banks are automating routine, low-value tasks using AI agents, allowing technical staff to shift focus to innovation and high-impact work. Leaders should identify repetitive workflows primed for AI delegation to boost team efficiency.
- Build over buy is the dominant AI strategy: One-third of banks are developing AI agents in-house to ensure tight integration, control, and regulatory alignment. Decision-makers should assess internal capabilities and prioritize AI development tailored to operational needs.
- Scale is the next major hurdle: While 80% of financial firms are piloting AI agents, only 10% have deployed at scale, highlighting a common gap between experimentation and execution. Executives need clear scaling roadmaps that address infrastructure and business alignment.
- Skills and compliance are the biggest blockers: Talent shortages and regulatory complexity are the top obstacles to AI adoption, cited by 92% and 96% of banks respectively. Leaders should invest in workforce development and cross-functional compliance strategies to accelerate adoption.
- AI agent adoption is going cross-industry: Companies like PepsiCo, Walmart, and Colgate-Palmolive are also deploying AI agents, enabled by cloud integration. Businesses should evaluate how AI and cloud can jointly support automation beyond IT functions.
- Long-term AI strategy drives meaningful ROI: Without a scalable plan and senior-level alignment, AI pilots remain isolated and under-optimized. Leaders should commit to a long-term AI vision that integrates agents into the broader business model for sustained value.
