How to leapfrog innovation in the age of AI.

Early in a conversation about artificial intelligence, Adeniyi Adebayo started talking about road graders.

He grew up in Shaki, a town in Oyo State near the border with Benin Republic. To build a road there, he said, you had to bring in an engineer from India, a man who arrived with instruments and measured the gradient of the ground. To a child watching, the work looked like a rare science. It was not. You think this is rocket science, he said. It is not rocket science.

That memory does a great deal of work. For most of modern African history, expertise had to be imported, and it was expensive. What has changed is not that machines became clever. It is that knowing things stopped being the bottleneck.

Adebayo is well placed to make the argument. He trained as a chemical engineer at Kazan National Research Technological University, took a master's at Skoltech, and studied at MIT. While still a student he founded ProctorEdu, a computer vision company that caught identity fraud in online examinations and grew into a business doing eight figures in annual sales. He became a venture partner at Starta VC, joined Yandex, and worked on the launch of Yango in Ghana, taking a product built for one market and rebuilding it for another. He is now Global Chief Business Officer at Yango Group and chief executive of Yango Africa, covering more than thirty markets, and he invests in energy, education, and mobility ventures across Africa, Latin America, and the Middle East.

The useful part of that history is not the seniority. It is that he has built technology, and then had to make people pay for it.

Bits are easy. Atoms are not.

His central claim inverts the usual anxiety. The AI that dominates headlines, meaning robots, self-driving cars, and mass physical automation, is in his reading substantially overhyped. The quieter story, the one about software, is underhyped.

“The digital world is bits. The physical world is atoms. Bits are easy, because it is zeros and ones. Atoms are several different kinds of chemical compounds, elements, space and time interaction. Those are things you cannot manipulate very quickly.”

He is blunt about why the error persists. The people leading the field are computer scientists, and they carry an intuition formed in an environment where anything can be rearranged instantly. A car that drives itself still has to be charged. Its tyres still wear out. When a passenger is sick in the back of one, somebody has to clean it. More than a decade passed between Google's first self-driving prototypes and anything resembling a commercial robotaxi service.

Then he put a question to the audience. How many new cars are sold in Nigeria each year? The guesses came back in the millions. By his count the real figure is closer to twenty five thousand, and under a hundred thousand once used imports are added, in a country of two hundred million people. If a technology as thoroughly solved as the motor car has spread that thinly in a century, the case for household robots arriving next year is weak. The economics finish the argument. A robot costs something in the order of a hundred and eighty thousand dollars. A worker costs a small fraction of that in a year, and where labour is young and plentiful, the machine loses.

Robotics will come, he thinks, but first to the societies that have no choice. Where populations are ageing and there are not enough people to put against the problem, machines go in instead. Young countries will wait.

What becomes scarce.

If knowledge now costs a few dollars a month, the binding constraint moves. It lands on the people willing to act.

“You will no longer be constrained by what you know.”

He means that practically rather than sentimentally. Europe is ageing. Africa is not. What the world is short of is energy and appetite, and that is the thing a young continent holds in surplus. The old dependency, importing a sixty year old expert to supervise work that was never truly esoteric, simply dissolves.

He is unsentimental about what this costs. AI is already displacing roles inside his own organisations, and he has had the conversation face to face with someone whose work a machine now does faster and more accurately. What he offered was not comfort but a redirection: take the time this frees, and do for many companies what you used to do for one. That person went away and started a business.

He also thinks the interface is about to change. Today people talk about prompt engineering, which is telling a system what to do. The skill he expects to matter is what he calls intent engineering, which is telling a system what you want to become and letting it derive the instructions. He described someone he knows in Amsterdam who has handed the running of several companies to agents, trained on his own writing so they answer in his voice, with access to his email and messages. The executives inside those companies have not noticed.

Why impressive products fail.

Asked what separates a product people adopt from one that merely looks good, he did not talk about technology at all. He talked about behaviour.

His example was a Lagos company running commuter buses on fixed routes, collecting workers in the morning and returning them at night. It worked because it fitted two genuinely predictable journeys. Then something the founders never designed for happened. Riders began forming communities on the buses, and then cooperatives inside those communities. People who could by then afford cars kept taking the bus, because the forty minute conversation on the way to work had become the reason to go.

Against that he set Uber's difficulties in Nigeria. The service was built on an assumption, that driving is casual work done between other things, which does not describe the market. In Lagos it is the whole job, all day, and the experience carries that weight.

“The critical thing is to think about the behaviour of the people using your product, and not just meet them at the point of their need but make it a delightful experience.”

He drew a hard line between products people return to and products people are stuck with. Nigerians lean on bank transfers and cash agents, he said, because the alternatives do not exist, not because anybody enjoys it. Necessity is not product market fit, and mistaking one for the other flatters a business right up to the moment somebody builds the thing people actually want.

Distribution is the moat now.

Until recently, the advice to build the best possible product was sound. He thinks the past two years ended that. Once a capable model can assemble a competent digital product, building stops being the thing that separates you, and quality converges.

What is left is whether you can put the product in front of people. His advice follows from it. Build in public, write, and gather an audience that trusts you, because that audience is the distribution. The technology will be copied. The relationship will not.

Patience, and the capital stack.

On the move from a working company to something that reorganises a sector, he was equally unromantic. Ambition at that scale is a financing question, and the common failure is choosing the wrong instrument for the game being played.

Equity suits businesses that can deliver venture returns. Infrastructure often cannot, and belongs on debt. He worked through battery swapping to show it. A commodity service charging a couple of dollars a swap will never throw off equity returns, but it can service development finance debt, provided the revenue is earned in the same currency as the loan. Take dollar debt against naira income, and a single devaluation ends the story.

Then there is time. He pointed at the Dangote refinery, roughly two decades from groundbreaking to production, financed through a stack of personal equity, bank debt, and international debt, with export volumes contracted early and boldly. If a builder with that much capital and access still needed twenty years, he asked, what is a realistic expectation for anybody else.

“Whether or not you spend the time building what you want to build, that time is still going to pass.”

The part that was not about technology.

He finished on constraints, and put the first one inside the builder's own head. Most problems have been solved somewhere by somebody, and are now learnable. The second constraint is impatience. The third is isolation, because the things worth building need collaborators, and the speed of one person is the wrong measure of anything.

He asked for flexibility about routes. A friend who wanted to build spacecraft did not take the expected path through an American university. He went from Nigeria to Russia to Germany and then to New York, where he now works at one of the largest space companies in the country. The destination held. Everything about how he reached it bent.

And he warned against a version of success that costs too much. He described older friends who built and ground and acquired, whose children are well travelled and do not much like them. For twelve years he has taken the same trip with the same group of friends, planned from January. He offered it without embarrassment as the reason the rest is worth doing.

“Life should be a lot more about an abundance mindset. But be very flexible about how you get there.”