How AI is becoming banking’s next big turning point

The first automated teller machine arrived with fanfare in a Barclays branch in 1967. People queued to press a few buttons, collect their cash, and marvel at a future where they no longer needed to wait for the teller's counter to open.

More than half a century later, banking stands at a similar moment. The machines have learned to listen, speak, predict, and decide. And the next leap now feels much larger than a metal box dispensing cash.

Across the world, banks are betting on artificial intelligence to reshape how money moves, how risks are monitored, and how customers experience finance.

In Bangladesh too, the shift is underway. It is uneven, but it is unmistakably gathering speed. From chatbots in mobile apps to data-driven fraud detection, AI is no longer an experiment. It is becoming part of the everyday banking experience.

This is the story of how that change is unfolding, why it matters, and what lies ahead.

A new kind of automation

If you have chatted with a virtual assistant in your banking app, you have already met AI. It now works behind the scenes in many places where customers barely notice it. It scans for suspicious transactions. It checks credit patterns. It helps sort documents, summarise phone calls, and guide investment research. It takes on the slow, repetitive work that once required many staff hours.

In Australia, the Commonwealth Bank says its AI tools have helped cut customer scam losses by half. In the United States, JPMorgan has built its own AI platform to assist staff across its business lines. And in Bangladesh, mobile banking apps are beginning to weave AI into the features customers use every day.

The most striking change is not that banks use AI, but how much they now depend on it. What started as support for simple back-office tasks is expanding into areas once thought to require human judgement.

The rise of app-based banking

The clearest sign of this shift can be seen in customer behaviour. Banking apps have become the main doorway for millions of users. Dutch-Bangla Bank's NexusPay, Islami Bank's CellFin, BRAC Bank's Astha, City Bank's Citytouch and Prime Bank's MyPrime have seen rapid growth, both in user numbers and transaction volume.

NexusPay alone now serves 8.6 million users. BRAC Bank's Astha app has crossed one million. Prime Bank's MyPrime app has introduced Bengali language support and SME banking, helping it post threefold growth in a year. Mutual Trust Bank's Neo app, which incorporates AI features and offers more than 200 services, has already driven a surge in digital account openings and deposits.

What customers see as convenience is powered, in many cases, by invisible AI. These systems verify identities through image recognition, recommend suitable products, translate content into multiple languages, and learn customer behaviour patterns to prevent fraud. Even something as simple as a reminder for a bill payment now rests on predictive models that study spending habits.

This digital shift is more than a matter of comfort. It changes the economics of banking. Fewer branch visits and quicker service reduce costs. Faster onboarding and automated risk checks widen access.

And as financial services expand beyond traditional bank counters, AI becomes the engine that keeps the system running smoothly.

From assistance to decision-making

Banks worldwide are now exploring what is called "agentic AI". Instead of only assisting employees, these systems can carry out tasks from start to finish. A major global bank tested AI teams that processed new customer applications independently, checking registries, verifying identities and completing compliance checks. Humans intervened only when cases deviated from the norm.

The productivity gains were striking. Basic automation might make a unit moderately faster. Giving AI full control could multiply output several times over.

While the idea sounds bold, it also raises some questions. What happens when a model makes a wrong decision? How does a bank explain a rejection or approval that came directly from an algorithm? These concerns are natural, and they are already influencing the pace of adoption.

However, the direction is clear: as technology advances, the line between support and autonomy will blur.

Opportunities for Bangladesh's banks

For Bangladesh, AI offers practical advantages. Banks can use machine-learning models to score credit for customers without formal credit histories. This is crucial in a market where many small businesses, informal workers and rural households remain outside traditional lending structures.

AI can also help strengthen fraud detection, especially as digital payments grow. With millions of transactions moving through mobile apps, real-time anomaly detection is essential. AI systems can flag unusual behaviour much faster than rule-based tools.

Operational efficiency is another major opportunity. Document-heavy processes like KYC checks, loan applications and compliance reporting can be automated. This reduces manual errors and frees staff to focus on advisory and relationship-oriented work.

Some banks are already experimenting. But studies show that most institutions in the country are still in early stages of AI readiness. Many lack cybersecurity frameworks for AI and have yet to include AI in their operational policies. This gap suggests that the sector is ripe for investment and innovation.

The regulator steps in

The Bangladesh Bank now plans to introduce a dedicated AI policy for the financial sector. A working team is drafting guidelines that focus on safe, ethical and regulated use of AI. The central bank is also considering building its own large language model to avoid the risks of sending sensitive data to foreign servers.

The policy aims to improve economic forecasting, analyse credit and market risks more accurately, and strengthen fraud detection. It also intends to help regulators present complex information in simpler language and respond more quickly to public complaints.

These steps reflect a realisation that the banking sector is changing too quickly for old supervisory tools. Responsible AI will require strong data governance, transparency, and protection against misuse. Banks and regulators will need to move together.

The human cost and the human need

The excitement around AI must also sit alongside a more sobering reality. Several banks overseas have learned that rapid automation can lead to job losses. When a major Australian bank replaced call centre roles with an AI chatbot, it later admitted that the transition lacked clarity and empathy.

The lesson is simple. Technology may change workflows, but banks still operate in a human environment. Customers want assurance. Staff need support and retraining. Responsible adoption will be as important as technical advancement.

The road ahead

The future of banking will not be shaped by AI alone, but by how banks choose to use it. They will need clear strategies, reliable data foundations, cybersecurity resilience, and teams that can manage and explain AI systems. They will also need to keep a human touch in areas where trust matters most.

What lies ahead is both exciting and demanding. Imagine applying for a loan late at night and receiving an approval within minutes. Imagine a banking app that understands your spending habits, warns you about risks, and suggests small changes that could improve your financial health. Imagine fraud systems smart enough to stop an attack before you even notice.

These possibilities are no longer distant ideas. They are approaching fast.

The change will not arrive with the drama of the first ATM. It will happen quietly, through apps that feel more intuitive, services that respond more quickly, and systems that learn from every interaction. But make no mistake. Banking is entering a new era, and the shift will touch every part of the industry.