Here's how it works. When you tap "call support" in the Capitec app, Pulse fires immediately. It pulls signals from across the bank's systems, including payment data, app diagnostics and risk controls to surface a contextual snapshot for the agent before they pick up. 

By the time the phone connects, the agent already knows whether you're calling about a declined transaction, a technical error or an account limit issue. You skip the part where you explain your problem from scratch.

Pulse is not a chatbot. It doesn't talk to customers. It's an AI co-pilot that works behind the scenes for the human agent, built by Capitec's internal software development team on Amazon Connect and AWS infrastructure.

The pilot results are specific and worth unpacking. Agents in the test group initially performed about 7% slower than the contact centre average. After Pulse was introduced, their call handling times dropped by 18%, which Capitec describes as a 26% net efficiency improvement when measured against their starting baseline.

That distinction matters. The 26% figure isn't a standalone improvement on an already-strong team. It's the swing from below-average to above-average, which suggests Pulse is particularly effective at lifting the performance floor. For a bank with 22 million clients and a contact centre handling significant volume, compressing every call by even a minute has material cost implications.

CIO Andrew Baker described the core engineering challenge as a physics problem: gathering data from across a massive retail banking footprint and presenting it in the seconds between a customer tapping "call" and an agent answering. That required building a new database architecture from scratch to deliver real-time contextual information as calls land.

Capitec claims Pulse is a world first. TechCentral's coverage noted that several major technology vendors already offer agent-assist platforms with broadly similar real-time context capabilities for contact centres. 

The distinction Capitec is drawing appears to be that Pulse delivers real-time situational understanding (pulling live signals from banking systems in the moment a call is initiated) rather than the more common approach of surfacing historical customer data and interaction logs.

Whether that distinction holds up under scrutiny is debatable. What's less debatable is that building this internally, rather than buying a vendor product, gives Capitec full control over how the system integrates with its banking stack. For a bank that has consistently differentiated on technology (and recently acquired fintech Walletdoc for R400 million), that control is strategic.

Pulse is event-triggered and time-limited. It activates only when a client contacts support and operates only during that interaction. It does not run in the background, does not perform behavioural profiling, and Capitec says it operates within the scope of its existing privacy policy.

That's a sensible design choice for a system that is, in effect, reading your banking behaviour in real time to predict why you're calling. The privacy architecture is worth noting precisely because it's done well. The system knows a lot about you in that moment, but it forgets you immediately after.

Most AI in banking lives in the back office: fraud detection, credit scoring, document processing. Pulse is unusual because it's a customer-facing AI that the customer never interacts with directly. The person on the other end of the call is still human. They're just a better-informed human.

For South Africa's broader fintech and banking ecosystem, this is a concrete example of AI creating measurable operational improvement rather than being deployed as a marketing talking point. The 18% call time reduction and 26% efficiency gain are auditable claims from a pilot with a control group, not projections from a pitch deck.

Whether other SA banks follow with similar tools will say a lot about how seriously the industry is investing in AI infrastructure versus AI announcements.

This news first appeared in our 11 March ‘26 edition on Bantu Stall team experiences.

For a look into the future of fintech in Africa, get yourself to our AI and fintech event featuring OpenAI, Paystack and Stripe in April 2026.

See how AI adoption is reshaping the continent in Africa AI adoption. Explore what South Africans are actually doing with AI in SA AI usage 2026. And read how Orca Fraud is using AI to fight payment fraud across 70 countries.