There is a website called RentaHuman.ai that exists to connect autonomous AI agents with human beings who can carry out physical tasks in the real world – things the agents cannot yet do themselves. Its tagline? “Clear briefs, no drama.”
It sounds like science-fiction. It is, in fact, very real. More than half a million people have already signed up, hoping to sell their services to machines.
For Morgan Goddard, partner and head of software engineering at South African management consulting and technology services firm iqbusiness, the existence of such a platform neatly encapsulates where we are in the AI revolution: a moment of such dizzying acceleration that the traditional relationship between humans and technology has begun, quietly but unmistakably, to invert.
“Maybe one day my AI bot will just do it for me,” Goddard says, laughing. “And we’ll just chill while the machines are doing all the work.”
He is not entirely joking.
“Humans have always been typing to machines in machine language so that we can get applications to work for us,” he tells TechCentral. “That’s kind of nonsensical. We should actually be talking English to computers, so that they can go and build things. That’s what large language models have done for us.”
The rise of so-called “vibe coding” – using natural language to generate functional software without writing a single line of code – has, in Goddard’s view, fundamentally disrupted the economics of software development. Tools like Lovable and Replit now allow a non-technical person to describe what they want, and receive a working web application complete with a database and a live URL, often within a few hours.
‘Honestly mind-blowing’
“Traditionally, you would need a designer, frontend developers, backend developers, DevOps people – possibly a four- or five-person team,” he says. “Now a non-technical person can fire that up in minutes. No syntax understood, no technical capability required.”
For experienced developers, the shift has been equally dramatic. Agentic coding assistants embedded in development environments now autocomplete, suggest, test and architect code – operating, Goddard says, at speeds that make conventional programming look archaic.
“These tools are speeding up your coding ability by something like 10 000 times,” he says. “And then, in the last few months, we’ve seen multi-agent orchestration come into the picture – an architect agent, a tester, a designer, a reviewer, a coder – all working simultaneously, autonomously, on your behalf. The amount of work these systems can do in a systematic way is honestly mind-blowing.”
Read: Dr Google, meet Dr Chatbot – neither is ready to see you now
The implications for the profession of software engineering are profound – and, Goddard acknowledges, somewhat unsettling.
“Team sizes will reduce,” he says. “What used to be a 10-person job is now potentially a two- or three-person job, or even a single person’s job. You’re becoming an orchestrator, a conductor. You’re allowing these tools to generate things on your behalf.”
Not all vibe coders are equal, he is quick to add. The more technical the operator, the better the output, because understanding what good code looks like remains essential to evaluating what the AI produces. “The genius will lie in whether you have the aptitude to understand if the output is of quality,” he says.
But this raises a troubling question about the future of the profession’s pipeline. Senior developers are battle-tested precisely because they have spent years writing code, making mistakes and understanding the nuances of their craft. If junior developers never acquire that grounding – because agents are doing the work from the outset – where do tomorrow’s senior developers come from?
“Do we still train people in syntax and the fundamentals before we allow them an agent?” Goddard asks. “Or do we just not care, because the agents are getting better and better? The universe of code that has been scraped by the most powerful language models on Earth – is that really worse than a general software developer? It knows every syntax, every language, every principle. It just needs the context.”
He believes the answer, in time, will be that it simply will not matter. “I think in the future we would just know it is done, and we wouldn’t care how it works. Like driving a car – you get in, you drive it, you don’t care how the engine works.”
The cybersecurity dimension of all this is one that keeps Goddard alert. The same tools accelerating legitimate software development are equally available to bad actors – and the threat landscape is shifting accordingly.
“If you have a coding agent cracking a system, and on the other side you have AI trying to stop it, both parties have the same tools,” he says. “It will escalate. It absolutely will escalate.”
More immediately, he is concerned about the risks posed not by sophisticated adversaries, but by ordinary users who do not understand what they are doing with the tools at their disposal.
Inequality
“People are working in four different agents simultaneously, pasting company information into them, and they have no real idea what is going on. Is the data being trained on? Is it being leaked? Companies need to take this extremely seriously – how they utilise these tools, how they train their people, the etiquette around using them.”
He likens it to the bring-your-own-device security nightmare that IT departments grappled with a decade ago. “But this is way worse,” he says. “Way, way worse.”
Then there is the bigger picture – the one that, Goddard admits, keeps him up at night.
Read: OpenAI unveils GPT-5, promising ‘on-demand software’ and PhD-level expertise
In South Africa, a country of extreme inequality and persistent unemployment, the displacement of routine cognitive and administrative work by AI agents is not an abstraction – it is an existential threat to millions of livelihoods.
“We see it already in retail,” he says. “You take one photograph of a garment, and using image models you can generate lifestyle photos, videos, social media posts, look books, brochures, all of it – from a single image. In that stream of work, we used to have photographers, designers, shoot locations, many people involved. Now we don’t. Apply that same thinking to admin, finance, accounting – every single one of them is impacted.”
He is sceptical of the techno-optimist argument that AI, like previous tech-led revolutions, will simply create new categories of work to replace those it destroys.
“Every other revolution happened at a pace that society could cope with,” he says. “This one is like jumping off a cliff. It’s fast, there’s enormous money behind it and it confronted us before we had time to figure it out. It took years to industrialise the car. With this, in about a year, you’re displaced. That’s different.”
So, what should a young person considering a career in software do? Goddard has a 10-year-old son and has thought about this question carefully.
Read: AI won’t replace software, says Nvidia CEO amid market rout
“There is no better time to be an entrepreneur in technology, because it’s completely wide open,” he says. “Take courses in entrepreneurship, finance, people management, marketing. Get real-world experience. But if you were planning to be the engine for writing code? That is not what you should do. Go further up the value chain.”
And the qualities that will matter most in whatever comes next?
“Curiosity, adaptability and resilience,” he says. – © 2026 NewsCentral Media
Get breaking news from TechCentral on WhatsApp. Sign up here.
