Kathy Gibson reports from the AI Everything Summit in Nairobi – Africa has no shortage of ideas, use cases and innovation when it comes to developing AI applications.
But having a great idea, or an elegant algorithm, doesn’t necessarily translate into a bankable product.
Getting from pilot to repeatable revenue is hard, says Winnie Mang’eni, founder and CEO of Pawa AI, US.
“We see a lot of African AI companies stuck in a perpetual loop of piloting without ever getting to production,” Mang’eni says.
But many of the ingredients for success are there, says Emanuele Spampinato, CEO of Harmonic Innovation Group, Italy. “There is passion capital and access to a huge market. But 99% of the world’s data centre capacity is outside of the continent. And too often the exit route for start-ups is acquisition by a multinational.”
Samwel Magesa, chief data officer at NBC Bank, Tanzania, agrees that data centre capacity is at the heart of our poor ability to scale.
“Africa is home to 20% of the world’s population – and it will be 40% by the end of the century. But we only have 1% of the data centre capacity. This is largely because we have a power problem. And if you don’t have power, you won’t have processing.”
Moving past the processing issues, Magesa points out that many innovators fail when it comes to scaling their systems.
“Startups tend to work well on prototype: the problem is with scale. Anyone can write a good algorithm, but AI relies heavily on data. I always ask what data was used to create the model because, if it’s not connected to local context it probably won’t be able to scale.”
Magesa urges developers to heed the 10-20-70 rule, spending 10% of their efforts on the algorithm, 20% on the technology, and 70% on people and processes.
Mang’eni believes the main challenge lies with the market. “The hardest commercial problem that startups have to contend with is finding a customer that is willing to buy their product,” she says. “We have a lot of talent in Africa and, while capital can be hard, it’s not impossible.
“Customers, though – especially African enterprises and governments – have long procurement cycles, with lots of governance. If you are building for the enterprise customer, this may be where you are blocked.
“Even if you get into procurement, they will typically look to foreign companies with proven track records rather than the startup,” Mang’eni says.
To ease this challenge, she thinks governments should create procurement processes that enable local providers.
“We need funders that are willing to take the product all the way; and enterprises that are willing to commit to local vendors.”
Magesa points out that the money flowing into Africa is very low, at just 1% of total global funding.
“So funding is a challenge, but it’s not the only issue,” he says. “The big issue with Africa is that we haven’t been able to figure out an ecosystem that works for Africa.
“We have our own unique problems which developers have to factor in. We have tried to adopt models, but in Africa we need to create our own ecosystem that works in our context.”
Collaboration will be key, Magesa adds. “If banking, government and venture capitalists are all involved; if we create the digital rails; and we all push the same agenda we can achieve what is needed.”
This would involve creating local context, with the right policies and accountability built in across the ecosystem, he explains.