Lloyd Matthew, Founder and CEO of Untapped AI.
There are two structural barriers that prevent global AI voice platforms from delivering in South Africa: accent authenticity and latency caused by international routing.
Untapped AI, the Cape Town-based company that identified these barriers, has solved for them, too. Rather than adapting offshore software, the company rebuilt the solution from the ground up for South African conditions.
Globally, almost all AI voice systems are trained on US and European datasets. When deployed locally, these limitations are clear. Mispronunciation of South African names and towns, robotic cadence and limited fluency across all 11 official languages were common to experience.
In voice automation, trust directly impacts uptake. If a voice sounds unnatural or foreign, engagement drops immediately.
“Authenticity shouldn’t be a feature,” says Lloyd Matthew, founder and CEO of Untapped AI. “It’s a fundamental foundation. So we recorded five South African AI voice agents in our studio. This way, we achieved contextual understanding, which ensures conversations feel natural and locally fluent, as well as the spoken words we need to build out a workable AI solution.”
The second challenge is infrastructure.
Most international AI voice providers process calls through overseas data centres. That means every local call travels thousands of kilometres before returning a response.
“In order to succeed we must have local storage in data centres here,” Matthew explains. “If the infrastructure isn’t local, the experience will never be.”
Untapped AI tackled the issue at infrastructure level by partnering with former telecommunications executives to build a carrier-grade voice infrastructure designed for national scale.
All AI infrastructure is hosted inside South African data centres in partnership with Nvidia, ensuring local-only routing and millisecond response timing.
With fully localised AI voice agents, direct local support and pricing in rands, it’s not unfair to say that this could see the breakthrough of voice automation in SA.
“There are two models. One is for the SME market and one caters to the enterprise corporate level,” says Matthew. “We’ve structured this so that both growing businesses and large corporates can deploy automation without compromise.”
Matthew believes the shift to AI-driven automation is not optional.
“Businesses that don’t scale and shift now will be lost. We’ve seen this in the supermarket space. One major player adapted and redefined the industry while the other stayed stagnant.”
The implication is clear: businesses have recognised the urgency to modernise the customer experience, even if they haven’t fully defined the solution.
As AI automation becomes the new competitive baseline, South Africa now has a homegrown infrastructure capable of competing on a global stage, without sounding imported.
