In a world where connectivity is critical, the strength of the network behind it matters. But equally important are the people with the expertise to build, manage and continuously evolve that network.
Meet our Experts puts the spotlight on the specialists and leaders behind Vox – the people whose technical knowledge, strategic thinking and industry experience play a fundamental role in how we deliver for our customers.
In our opening edition, we speak to Keith Laaks, Executive Head of Technology at Vox. With extensive experience in the telecommunications industry, Keith operates at the heart of one of the most critical areas of our business. We unpack his journey, the realities of managing network infrastructure at scale, the technologies reshaping connectivity, and what it takes to build networks that can keep pace with the demands of an increasingly connected world.
Every now and then, you come across a true industry titan, someone whose insight, knowledge and expertise makes you say “whoa”. This is one of those moments. Read our full Q&A with Keith below, and join us in appreciating a true connoisseur of his craft.
Tell us a little about yourself and your role as Executive Head of Technology at Vox.
I matriculated in Mthatha (formerly Umtata) and went on to complete B.Eng and M.Eng degrees at Stellenbosch University. I then completed compulsory military service as a signals officer, before being seconded to the CSIR, which was running a project related to my thesis. I joined Telkom next, working first in its R&D environment and then in the Technology Strategy division, on several projects that gave me a foundation I have built on throughout my career. From there I worked at businesses such as Siemens Business Solutions, Origin and Sentech, before joining Orion to help build one of South Africa’s first VoIP networks. Orion was acquired by Datapro, which later became Vox and I have been here ever since.
Today I lead Vox Core, the part of the Vox organisation responsible for designing, operating and maintaining the network infrastructure and OSS applications that deliver our broadband connectivity, voice, email and hosting services.
How did your journey into technology begin, and what ultimately led you to where you are today?
I was exposed to technology from an early age, influenced by my father and grandfather, who were both ‘radio amateurs’ – or ‘hams’, as they are commonly known. Growing up, we always seemed to have cutting-edge technology around the house. While I was living in Mthatha as a teenager, the Citizen Band Radio craze took off, and my friends and I had great fun talking to people all over the world. Remember, these were the days before cell-phones, so being able to talk to people around the world ‘for free’ was intriguing. We were also one of the first families to buy our own ‘home computer’, an Apple ][, which we quickly connected to the radios to access ‘bulletin boards’ – the early ancestors of the internet. All of these influences led me to choose electronic engineering and telecoms as a career, and brought me to where I am today.
Was there a particular technology, project or moment early in your career that made you realise, “This is what I want to do”?
It was less a moment than a gradual realisation, and it happened at Telkom. Moving from the R&D environment into Technology Strategy forced me to think about why a technology mattered commercially, not just whether it worked. Being in the room where those choices were made, and then seeing the consequences play out across a national network, was when I realised that this intersection of engineering and strategy was the work I wanted to do.
Technology changes incredibly quickly. How do you personally stay ahead of what matters without getting distracted by every new trend?
I don’t try to follow every little thing happening out there. I just stick to a solid, hand-picked mix of sources. I tune into podcasts, hang out on the right Subreddits, and watch YouTube channels on stuff I actually care about. Plus, I subscribe to the main creators and groups in those spaces so the good material comes straight to me instead of me chasing it down.
The real trick is staying disciplined with your setup: by picking your sources properly and unsubscribing the second something goes stale, you keep the signal high and cut out all the noise. At the end of the day, having real depth in a few key areas beats a cheap, surface-level know-it-all vibe on every new trend every single time.
For me, the biggest game-changer isn’t even one specific tech tool. It’s the sheer speed of it all.
Technology is moving way faster than our institutions can keep up, and that gap is just getting bigger, not smaller. Companies are still out here setting three- and five-year business plans while the tech they’re planning around changes every couple of months. So by the time the board actually approves a strategy, the ground has already shifted underneath them.
I don’t think people are being stupid or lazy here; it is just that humans are not wired to handle this kind of pace. We evolved to think in straight lines and expect things to move slowly, just like they did last year. Exponential growth isn’t in our DNA, so we constantly sleep on how fast something goes from a cool gimmick to an absolute necessity if you want to stay in the game.
The businesses actually taking this seriously are not just buying new gear, they are changing how they make decisions. We’re talking shorter planning cycles, constantly running small experiments, and having the discipline to check their assumptions long before the paint even dries.
AI dominates almost every technology conversation today. What do you believe is genuinely transformative about it — and what is simply hype?
What’s genuinely wild is how fast standard foundation models have turned into real ‘agentic AI’ systems. We’re talking about models that can reason, use tools, plan out multi-step tasks, and do it all with barely any human hand-holding. Add in the multimodal and robotics stuff these systems are starting to pull off, and it’s obvious this isn’t just a software update with extra steps.
It basically makes high-level thinking cheap and available to everyone. This flips the economics of knowledge work, software engineering, science, and eventually physical labour too.
For me, the hype around all this boils down to three things.
First, the terminology is way overblown. Calling basic statistical models ‘intelligence’ and slapping the word ‘agent’ on what’s often just a thin wrapper around someone else’s model.
Second, there’s this assumption that rolling it out is easy and cheap. Sure, a demo can look slick, but actually fixing your data quality, integrating it into your systems, sorting out governance, and getting people to change how they work? That’s slow and it’s not cheap.
Third (this is the big one) is the idea that this stuff removes the need for human judgment. It doesn’t. If anything, good judgment matters more now, because confidently acting on an answer that sounds right but isn’t costs you a lot more than it used to.
My rule of thumb is simple: if a vendor can’t tell you straight what happens when the model gets it wrong, they’re selling hype, not a real product.
From connectivity and cloud to cybersecurity and AI, businesses are becoming completely dependent on technology. Where do you think the biggest vulnerability now lies?
Technology adoption is accelerating at speed, but risk management, compliance, and institutional decision-making are still moving in low gear. This mismatch is driving a serious escalation in shadow IT: non-technical staff can now use no-code platforms like Lovable to prompt full-stack applications into existence in a matter of minutes.
When these ad-hoc setups handle Personally Identifiable Information (PII) without proper internal oversight, Row-Level Security, or basic access controls, formal governance gets bypassed completely. This leaves the business wide open to major data leaks and regulatory non-compliance risks.
Looking specifically at South Africa, what is the biggest technological challenge we need to solve over the next five to ten years?
The biggest technological challenge South Africa faces over the next decade is not the adoption of AI, cloud computing, or automation, but the fragile physical and institutional foundation on which all digital systems depend. Reliable electricity, functional municipal governance, secure infrastructure, and basic law and order are the silent prerequisites of every tech stack. Without them, even the most ambitious digital strategies fail when substations are sabotaged, cable thieves strike, or local councils collapse under corruption and debt.
Furthermore, as municipal services degrade into chronic water cuts, crumbling infrastructure, and high violent crime, the globally mobile engineering and cybersecurity talent required to build South Africa’s digital future simply leaves for foreign markets.
Compounding this is a severe tax handicap: a massively inefficient government and state monopolies that spend a disproportionate share of GDP compared with other developing and developed countries against which our businesses must compete. The resulting ever-increasing tax burden means businesses cannot grow because they struggle to compete internationally, while people with capital prefer to invest in countries offering more favourable tax and business environments. Job creation in South Africa is stifled as a direct result.
As the government spends more than it collects in taxes, debt rises and it causes inflation, which is the hidden tax that few recognise.
Solving these foundational, last-mile bottlenecks of energy resilience, infrastructure protection, local government efficiency, and a more competitive fiscal environment is the single highest-leverage technological priority for the decade ahead, because every smart city, public sector deployment, and AI initiative ultimately rests on that foundation.
AI stepping out of the screen and into space, will be the big story. The “regular AI” that we are mostly exposed to today, has changed how we think and communicate and people are obsessed with chatbots and image, video, music generators. But the thing that is really going to fundamentally change things over the next decade, is the “Humanoid AI Robot” that can move through and act in the physical world.
Robotics used to mean clunky single-purpose arms bolted down in a safety cage, like in a car manufacturer’s assembly line. What’s happening now, mostly under the radar, is that AI models capable of seeing, reasoning, and acting (the physical-world counterparts of chatbots) are merging with far more flexible hardware. Tesla’s Optimus, along with Chinese platforms such as Figure’s humanoid robots and Unitree’s G1 and H1, demonstrate the shift. These products can operate in unstructured human environments rather than being locked into a single repetitive industrial motion.
Robots are adapting to us, not the other way around. Old-school automation meant redesigning a whole workspace around the machine. These general-purpose AI enabled robots changes everything. They are built to work in the world as it already exists. Climbing stairs, opening doors, picking up a power tool, navigating a cluttered warehouse or an old factory.
The same way AI has dramatically reduced the cost of writing and coding, these robots will do the same for physical routine work, such as stocking shelves, sorting packages, cleaning, harvesting crops, general maintenance. Physical labour will get a lot cheaper. That changes the math in many industries, especially the labour intensive ones.
Abroad, aging populations and labour shortages in logistics, healthcare, construction, and manufacturing are major drivers accelerating the development of more capable (and cheaper) robots. In South Africa the demographic picture is different, yet the commercial logic still points strongly toward automation. Chronic skills shortages, combined with periodic labour strikes and protest actions that disrupt operations, mean many businesses, especially those that require consistent, uninterrupted output, will tend to pursue full or augmented robotic solutions.
When the cost of owning and operating these systems falls below the total cost of employing and managing human labour, it will make the economic case for transitioning to general-purpose robots even more compelling.
Home and care work changes too. As robots get better at handling delicate, tricky tasks, such as folding laundry, helping an elderly relative safely, a massive chunk of unpaid household labour could get offloaded to these robots. And if the arrival of domestic appliances is any guide, this shift will happen even faster.
Finally, nerd question: if money, time and commercial viability weren’t considerations, what technology or problem would you personally love to spend your time working on?
I’d want to build analogue AI inference chips. The kind that do matrix math with actual physics (current, voltage, resistance) instead of a GPU sweating through billions of digital multiply-adds.
Our brain operates on just 20 watts (barely enough power to light a single dim bulb) yet it effortlessly recognizes faces, holds complex conversations, and controls every physical movement in real time. Meanwhile my GPU pulls 400 watts and needs its own cooling system just to tell me my cat photo is, in fact, a cat.
We built the most inefficient possible way to do math: flipping bits back and forth at gigahertz speeds to simulate an analogue process. And now we have to build nuclear power plants to power it all. Somewhere, thousands of transistors are executing a high-stakes, million-cycle dance just to flip a single digital bit, when a tiny trickle of raw current could have carried the signal directly.
So if you give me infinite time and no commercial pressure, and I’d love to help make AI run on the energy budget of a light bulb instead of a small country.
So ends episode one, with plenty to think about and a lot to marvel at. Keep your eyes fixed firmly on our socials for our next edition, and be sure to get in touch should you need any help from the wealth of experts here at Vox.
About Author
