Agentic AI reality check: don't confuse hype with readiness
By Jonathan Tanner, Senior Director – Financial Services & Insurance, APAC, Pegasystems
Wednesday, 02 July, 2025
Agentic AI is no longer a speculative concept in innovation labs or a buzzword in boardrooms, it’s becoming a transformative force in business and central on strategic agendas.
In the push for efficiency and scale, agentic AI and its ability to make independent decisions and actions is swiftly being viewed as the defining factor between industry leaders and those left trailing behind. In banking, the technology represents a seismic shift for the industry, from advanced fraud detection to customer service automation, faster document processing and regulatory compliance support.
So, it’s unsurprising that leaders are eager to push on, but the banking industry’s preparedness lags behind.
Laying the groundwork for AI integration
Banking execs are right to champion agentic AI, but in a bid to act quickly they are overlooking one crucial issue: the legacy systems that form the core of the institutions themselves.
For AI agents to function effectively, they need high-quality, connected data, modern integrated technological components and strong cybersecurity. Instead, most banking systems still struggle with siloed data, rigid monolithic systems and outdated defences, leaving them inflexible, vulnerable and stuck in the past.
For IT leaders, this presents both challenges and opportunities for fundamental business impact — to integrate agentic AI effectively and harness its full potential. To do so, we must lay the groundwork first, modernising the core infrastructure, ensuring seamless data integration, real-time processing capabilities and robust compliance frameworks. The transformation is not merely a technical upgrade but should underpin banking strategies to pave the way for the future of the industry. It is critical therefore, that IT experts and banking leaders work hand-in-hand to proactively address the associated challenges that come with implementing agentic AI.
Despite the lack of system readiness, financial institutions are pushing forward, piloting use cases such as autonomous portfolio advisors, independent fraud detection systems and hyper-personalised customer engagement engines. These examples demonstrate agentic AI’s potential on a small scale, but widespread adoption remains limited as the technology and infrastructure struggle to catch up.
And that’s just half the battle. In fact, once the groundwork is laid for such powerful automation, we must consider the question of control.
Trust in AI hinges on human judgment
Once AI begins to make unsupervised decisions that impact customers, regulators and executives will demand heightened scrutiny.
This means human oversight remains essential to ensure AI complies with evolving regulation. For example, institutions will need to ensure that final approvals on AI recommendations that impact areas of regulatory oversight are underpinned by a clearly defined set of business rules, and will need to maintain robust audit trails to maintain transparency.
However you choose to approach it, the principle stands firm: AI governance must guarantee accountability, traceability and regulatory compliance. For this to happen, human judgment is not optional — it’s critical.
AI success is nothing without organisational alignment
In my experience, real success with new technologies can only be achieved when business, risk and IT are fully aligned, providing strong tech infrastructure, implementing secure compliance measures and skilling their employees effectively. In the case of agentic AI, this means moving beyond APIs to event-driven architectures for fast, contextual data; embedding strict governance with explainability, bias checks and compliance in workflows; and building teams that are fluent in both the tech and domain knowledge.
Ultimately, the biggest barrier to long-term success of agentic AI in the banking industry isn’t in the capability of the technology — it’s in the outdated systems, rigid processes and leadership mindsets that refuse to fully embrace new autonomous approaches. For IT leaders, the opportunity lies not in chasing the next breakthrough demo but in reshaping the infrastructure, governance and skilling that supports autonomy in the long term.
Successful businesses won’t be the first to simply deploy agentic AI, they will be the ones that can operationalise it effectively for scalability.
So, does your business have the solid foundations to support agentic AI? If not, consider what’s stopping you: the technology, or the organisation behind it?
Shadow AI exposes the cost of slow governance
Shadow AI is what happens when governance fails to keep pace with experimentation,...
How agentic AI will revolutionise customer experience in Australia
Agentic AI is changing the way that Australian businesses are adopting AI technologies.
How to prepare your data for AI success
Proving the AI-readiness of data is a process and practice based on the availability of metadata...