How AI is teaching banks to read customers’ mind

It could recommend home insurance the moment you complete the purchase, flag early signs of financial stress before you miss a loan repayment, and spot suspicious transactions within seconds of fraud occurring.

This is the future that technology firms and financial companies are increasingly building toward as artificial intelligence (AI) moves from experimental pilots into core banking and insurance operations.

In Kenya, banks, insurers, mobile money providers and fintech firms already generate large amounts of customer data, including deposit, withdrawal and fund transfer details, spending habits, budgeting trends, and audio recordings of customer service calls.

These firms are always working toward building tools that help them analyse the pools of data for valuable insights. Over the past decade, banks and Saccos have invested heavily in digital channels, moving customers from traditional banking halls to mobile apps and online platforms.

But while much of the public discussion around AI has focused on efficiency gains and job automation, banking and insurance industry executives are seeing a bigger prize in data intelligence.

Global IT firms like Salesforce, NTT Data, Oracle and Microsoft have built tools that use AI, data and automation to cut costs, understand customers better, sell more products and improve decision-making.

“The real value is when we actually look at the top line. How are we using AI to cross-sell and up-sell products?” Lauren Wortmann, NTT Data’s managing director of application services for the Middle East and Africa, told the Business Daily. “How are we using AI to identify inefficiencies in the revenue lifecycle and optimise those?"

Cross-selling refers to encouraging an existing customer to buy related products with their initial selection, while up-selling is encouraging them to purchase a more expensive version of the item they are already considering.

The concept is often described as the "next best action" — using customer data and predictive analytics to identify what a customer is most likely to need next.

If a customer begins exhibiting signals associated with buying a home, for example, AI systems could recommend a mortgage, insurance cover and other related products. The same principle could apply across savings, investments, lending and insurance.

Kenya's highly interconnected financial ecosystem makes such opportunities attractive to these technology firms.

Consumers regularly move money between banks, mobile money platforms, businesspeople, insurers and fintech applications, creating a complex web of customer interactions.

"You are walking a customer through a connected journey that doesn't just look at a singular bank ecosystem, but also looks at the partners within that ecosystem, such as mobile money," Ms Wortmann said.

The ability to build a single customer view is also attracting attention as lenders seek new ways to manage rising credit risk.

Nick Christodoulou, Africa’s vice-president for Salesforce, argues that most financial institutions already have large volumes of customer information but struggle to use it.

"My bank knows me better than I know myself because I have so many different touch points with my bank and any kind of related organisations that connect into my bank," he said.

"What's lacking is our ability to create one view of the customer and to get close to that customer, understand the thought process of that customer, but also have enough predictive analytics to actually almost pre-empt the move of the customer."

Kenyan banks reckon that such predictive capability could eventually help them identify signs of financial distress before borrowers default, improving lending decisions and reducing losses.

Beyond lending, AI is increasingly being deployed to combat fraud. Banks are experimenting with systems that can analyse suspicious activity across multiple platforms, automate investigations and accelerate responses when customers report stolen funds.

According to a 2025 survey by the Central Bank of Kenya (CBK), banks are using anomaly-detection models to identify unusual network activity, automate threat triage, and speed up incident response.

“The top three applications of AI and ML by institutions that had adopted AI were credit risk assessment at 65 percent, cybersecurity at 54 percent, and customer service at 43 percent. This was followed by e-KYC (know your customer) at 41 percent and fraud risk management at 40 percent,” the CBK said.

But amid concerns of massive job losses with increased automation, executives insist the technology is more likely to augment employees than replace them outright.

Customer service agents, relationship managers, underwriters and claims processors are increasingly being equipped with AI tools that surface information, recommend actions and automate repetitive tasks.

"What I'm seeing is more of a productivity and enabling humans to create better experiences, make better decisions and have more data and information to guide an organisation," Ms Wortmann said.

Aside from the industry's enthusiasm, local financial firms are cautious, spending significant time establishing governance frameworks on data privacy, security and responsible AI use before deploying the technology at scale.

Experts say regulators will play an important role in defining those boundaries as automation expands across customer onboarding, claims processing, fraud management, lending decisions and personalised product recommendations.

→ [email protected]

Follow our WhatsApp channel for the latest business and markets updates.