At the centre of this shift is Darren Bak, Head of Intelligent Data at Synthesis, who says it plainly: “We’ve watched the banking space evolve. We’ve grown with banks — and helped them grow in return.”

Traditional banks were once protected by regulation and scale. But today, they’re up against fast, digital-first challengers and customers who expect more — and compare their banking experiences not with other banks, but with Apple and Google.

Tech Giants Set the Pace

Apple Pay works because it’s invisible. Secure, seamless, and simple. Google, meanwhile, is going further — pushing into Agentic AI. At its recent I/O developer conference, Google showed how AI can guide users through product discovery and payment in real time. No standalone banking app needed — just search, click, and pay, all under AI supervision.

This isn’t just convenience. It’s a warning shot to legacy banks: Innovate or be outpaced.

Banks Need More Than Co-Pilots

While many institutions are still experimenting with chatbots and AI co-pilots, Bak believes the real power lies in domain-specific AI agents. “These agents,” he explains, “don’t just generate text. They solve specific, high-value banking problems using deeply contextual data.”

But there’s a catch: AI is only as smart as the data it feeds on. And in banking, data can’t just be available — it needs to be secure, structured, and ready to move in milliseconds.

That’s where Synthesis shines.

Infrastructure That Matters

Synthesis, the recipient of the AWS GenAI Prize of Excellence Award and voted the preferred software development company in South Africa for large business projects, has built resilient, distributed systems that prioritise data sovereignty, security, and scale. Its AI Launchpad platform allows banks to deploy advanced AI safely — with embedded security, real-time monitoring, automation, and cost controls that keep the CFO happy.

In high-stakes use cases like fraud detection, timing is everything. “Milliseconds matter,” says Bak. “And that’s the difference between stopping a breach or explaining one.”

Real-World Proof

If you want a glimpse into what’s possible, look no further than Vitality Group. With over 60 million life years of data, 4+ petabytes of behaviour-linked records, and 5+ billion device readings a year, they’re setting the global benchmark for AI in insurance. And they’re doing it from here.

“South African engineers are world-class,” says Bak. “We’re not behind. The only thing slowing us down is outdated systems and data silos.”

Start Smart, Scale Fast

So how should banks respond? According to Bak, the approach is simple:
“Think big, start small, and get to value fast.”

This means identifying high-impact use cases, building foundational data platforms, and delivering quick wins. One area where Synthesis has seen major traction: customer personalisation. Through rich customer data platforms, banks can serve up real-time nudges, contextual product offers, and frictionless experiences that actually reflect the customer journey.

The Stakes Have Changed

Switching banks is now easier than ever. Loyalty isn’t a given — it’s earned through smarter, faster, more personalised service. If banks want to keep their customers (let alone attract new ones), they need to stop reacting and start building.

The post We Built the Bank of the Future: Darren Bak appeared first on Synthesis.

How Do We Feel About AI Now?

The perception of AI depends on an individual’s perspective of AI, says Marais Neethling,  Principal for Artificial Intelligence at Synthesis Software Technologies. Some I might view AI as the next best thing to revolutionise the world, and others as a mechanism for taking jobs away and making people lazy.

Any negative connotation can be removed by simply changing perspective. If the perspective is to focus on the negative, that will be your belief. The reality is that with any disruptive piece of technology, there will be good and bad that come from it. An increasing number of people can now easily recognize AI-generated content, raising concerns about its use without proper disclosure. When should you use it? How should you use it? Is it okay to use it?

In the same way, the value extracted from AI largely depends on an individual’s intentions and motivations. When used thoughtfully to enhance or refine your work, it can become an incredibly powerful tool. If it is used with malicious intent, that could be a different conversation. All we know for sure is that it is here, and it is here to stay, explains Neethling.

Neethling anticipates seeing five AI trends in 2025.

1. AI Agents and AI Capabilities Will Become More Pervasive

In the day-to-day lives of consumers, the adoption of AI learning and understanding comes from the need of individuals to upgrade devices. Apple, for example, recently launched Apple Intelligence which is now available on the iPhone 16. This upgraded version of Siri can perform more tasks on your phone than the existing assistant. Phone and PC manufacturers are also bringing AI straight into the desktop, enabling users to access and learn how to use it.

2. Multimodal Advances in AI

The nature of the latest models will provide different modes of media consumption. Video, images and audio. Imagine an assistant that looks at your computer screen in real-time, all the time, and can make suggestions or automate tasks while you work. Depending on what is being worked on a, the tool suggests options, logs time against tasks, and automatically tracks it. This takes away the grudging admin of your day by automating it.

3. Impact of AI on Energy Consumption

For large enterprises and companies, AI models require a huge amount of computing to run, both during training and execution. Massive data centers are required which results in the need for enormous amounts of power to function effectively.  Understanding the amount of power and electricity needed is the responsibility of all individuals to be environmentally conscious. Changes to the economy, renewables, and other types of energy sources are expected with the high demand for energy supply needed to support AI transformation, first in the AI hubs around the world, and later in South Africa.

4. Virtual Reality (VR) Advancing into Wearable Tech

VR headsets are big and bulky; however, companies are now starting to launch smaller and smaller ones. Meta and other companies are working on reducing the size of wearable computers to fit into spectacles or sunglasses. AI can be integrated into these augmented reality (AR) devices to understand what the wearer is doing in the real world. The AI should be able to identify objects and people in the wearer’s field of view and supply interesting information about the subjects in real-time. It is a bit futuristic and maybe a little creepy but has the potential to unlock new opportunities as it evolves. It’s unlikely that we will see some of this in 2025, but it is definitely something that is on the horizon.

5. Industry Use Cases for AI

The medical and healthcare industry leverages AI to provide improved patient care, from accurate diagnoses to supporting patients in managing symptoms. Multi-modality introduces an exciting new dimension, particularly in analysing X-rays and other sensory inputs. Multimodal models enhance diagnostic capabilities by integrating diverse data types, offering richer insights compared to traditional methods that relied solely on textual inputs. Although medical procedures are regulated, it does not stop individuals from getting a second opinion and using AI to help diagnose patients could prove to be accurate.

There are many more use cases for AI across industries covering many topics to improve security, automate manual processes, create content, and develop innovations. This is just the start, I can’t wait to see what’s next, concludes Neethling.

The post The Latest View on the Horizon of AI – 5 Things to Expect in 2025 appeared first on Synthesis.