For years, digital banking has been measured by access. Ensuring clients can open an account without visiting a branch, transact from a phone, or get help outside normal banking hours shaped the first wave of banking digitisation and changed how millions of South Africans manage their money.
However, the next phase of digital banking will be measured less by access and more by the level of digital intelligence banks can offer. This implies far more advanced questions around how well a bank understands each client’s context, uses data responsibly, anticipates needs and helps people make better financial decisions.
While digitisation gave clients more control over when and where they bank. AI is changing the quality of that control by making banking platforms more adaptive, contextual and responsive. A client who needs help with a digital task expects a quicker route to the answer. A client reviewing spending or budgeting patterns needs deeper insight that reflects their actual behaviour. This means that AI will increasingly become part of the way clients interact with their banks. Clients are already used to digital assistants in search, shopping, travel, work and entertainment. Banking will follow the same pattern as AI-powered tools help clients interpret information, compare options and complete tasks.
Artificial intelligence is already quickly moving banking in this more personalised and impactful direction. Across financial services, AI is being used to improve risk management, strengthen fraud detection, support service channels and personalise engagement. The challenge is to ensure that this personalisation does not become intrusive. AI in banking should reduce complexity, improve relevance and support better financial decision-making. If it merely serves to create more prompts, offers and automated interventions, it will add noise rather than value.
To ensure this, the definition of the customer needs to evolve. While they will always be human beings with financial goals, responsibilities, preferences and rights, they are increasingly being supported by AI ‘assistants’ that help them gather information, assess choices and interact with service providers. This creates a new service design challenge for banks. Where, historically, they have built journeys around direct human interaction – even when those journeys moved into digital channels – they will now need to design for clients whose lives are augmented by AI and, over time, for non-human agents authorised to act within defined boundaries on a client’s behalf.
Agentic AI is driving this shift from automation toward autonomy in financial services, and that is reshaping understanding around personalised advice, customer interaction and financial decision support, while at the same time raising governance, fairness and risk questions.
This is a future that has to be approached very carefully. Financial decisions affect access to credit, savings behaviour, insurance choices and long-term financial wellbeing. As AI tools become more active in decision-making, banks will need to be clear about identity, consent, traceability and accountability.
At Nedbank, we have taken a deliberate and measured approach to this future banking reality. AI is already embedded in how we serve clients, and we are progressively building towards more autonomous capabilities in controlled contexts. That means making AI a part of core journeys and decision layers, but always keeping trust central to the client experience.
We see this trust as central to any bank’s effectiveness in an AI-informed future. That future will be shaped by whether clients trust the systems that use it. And that trust will need to be built on foundations of relevance, reliability, consent and clear accountability.
The opportunity for banks, therefore, is to use AI to make banking more personal without making it opaque, and to help clients make better financial decisions without removing their agency. This means serving people directly, while also developing the intelligent systems required by AI agents that act with their permission. The banks that succeed will be the ones that design for both audiences-the people banking with them and the AI agents banking for them.
