It has gone mostly under the radar, but the use of artificial intelligence (AI) by African tech startups is on the rise, with the sector becoming bigger by the week and attracting more funding.
Back in April, Google announced it was opening an AI research centre in Ghana, bringing the potential of Africa as a hub for AI and machine learning into sharp focus. Yet things had been bubbling along for a while before that, with a number of startups using these technologies in a host of different spaces.
Some of these have already secured VC funding, but investment into AI startups, or companies looking to utilise AI within their existing operations, looks set to grow substantially with the recent news that South African private equity investment firm Ethos has launched a ZAR600 million (US$42.9 million) fund for startups that will benefit from AI-based algorithmic decision making.
In fact, South Africa has already established itself as something of a hub for African AI. Several startups are making names for themselves in different areas, believing they have spotted a major gap and that increased uptake of AI and machine learning across multiple industries is inevitable.
AI for manufacturing
The Cape Town-based DataProphet is one of the main players. Managing director Frans Cronje has an MSc in statistics from the University of Cape Town, and began his working life consulting on machine learning for various companies as he identified business problems that could be solved with these advanced technologies.
Founded in 2015, DataProphet uses AI within the manufacturing sector, to improve the efficiency of the process by optimising the variable process parameters.
“Over the years, manufacturing businesses have amassed a lot of data which is underutilised, in some cases, only collected for compliance purposes. Most manufacturers are still using traditional approaches to analyse their data, which were not built to handle the amount of data being collected nowadays, therefore sub-optimal,” Cronje told Disrupt Africa.
“Much investment has gone into the setting up of infrastructure for the acquisition and storage of such data, with very little advanced analytics happening on it. You’ll, therefore, find that manufacturers are sitting on a lot of insights into their process within that data and the potential for a return on their investment.”
DataProphet developed a system specifically for manufacturing firms that helps them reduce defects and increase yield, and works with customers that include car manufacturers like Mercedes-Benz and BMW, and various foundries. Earlier this year, it raised a multi-million dollar funding round from South African VC firm Knife Capital to accelerate its global expansion.
Cronje says the nascent nature of the AI space means it offers huge opportunities to startups entering the market at this time.
“In general, AI has been rapidly introduced to processes around the world – and so it is underutilised. In Africa, that is also the case, more so as processes are typically less mature due to it being a developing continent, and skills are more scarce here,” he said.
“The extent to which AI can be used is so large that we have not even scratched the surface, and this is increasing with further developments in this field, moving into the deep learning space.”
More and more industries are “getting their feet wet” when it comes to AI, however, and more budgets are being made available.
“AI projects are still seen as innovation projects as most organisations are not sure of the results. Currently, AI is used to optimise processes – make them more efficient – but it does require the process to be in place. This will, however, gradually change as processes are built with AI in mind, rather than AI attached. So we will see a move from a manufacturer improved by AI, to a manufacturer built around AI.”
AI for agriculture
Another massive African industry with the potential to be hugely disrupted by AI is agriculture. Another Cape Town-based startup is taking the lead in this respect. Aerobotics has developed proprietary AI to process data it gathers using satellite imagery and drones, in order to learn about different tree and vine crops, analyse trees down to the canopy level, and provide insights and data to farmers that they would not have identified with the naked eye or through satellite imagery alone.
The startup, which raised funding of around US$2 million earlier this year and recently released a host of new products, has more than eight million trees in its database, which chief executive officer (CEO) James Paterson told Disrupt Africa means its AI is getting smarter and more accurate by the minute.
“Through our AI, we are getting highly accurate tree counts and size and health measurements on a per tree basis that help the farmer identify pests and diseases early and equip them with critical insights, so they can make better decisions to reduce loss and increase yield,” he said.
Paterson, who was himself brought up on a South African farm and therefore knows a thing or two about the challenges faced by farmers, said he fully believes in the potential of AI to positively impact farming and other industries around the world.
“AI is already being used in places like customer service, transportation, shipping and healthcare, but there has been a relatively lower uptake in farming up until the last couple years. The farming digital transformation is happening, and we believe we are at its forefront in Africa and around the world,” he said.
“The more we speak with farmers and partners in the agriculture industry, the more we are able to showcase how we are helping the agriculture industry, which is causing us to see an increase in the utilisation of our AI in the agriculture industry in South Africa and around the world.”
He believes the space will grow exponentially as the market matures, meaning startups already active in the AI sector have much to gain.
“There will be an increase in the use of the tech and a distinction between the different offerings out there, which will help Aerobotics immensely as we provide world leading early problem detection technology,” said Paterson.
AI for finance
Fintech is one of the major sub-sectors of the African tech space, and here too AI is having an impact. Leading the way is Nigerian company Mines, which uses AI to power its Credit-as-a-Service digital platform, enabling institutions in emerging markets to offer credit products to their customers with no smartphone required.
“Leveraging their own data sets, domestic institutions are able to serve loans to customers ignored by available credit systems and open up entirely new revenue opportunities. By mining high-volume data like phone records, bank records, and payment transactions in real-time, Mines can instantly assess credit risk in markets that lack robust credit bureau infrastructure,” said the company’s Nigeria managing director Adia Sowho.
“It then integrates its risk models with identity, origination, payments, loan lifecycle management, and customer service to form a holistic platform. The net result is a seamless user experience where partners’ customers can apply for and receive a loan in less than 60 seconds or make instant purchases with virtual or physical credit cards.”
Such is the impact of this that Mines has proven very attractive to investors, bagging a US$13 million funding round earlier this year. This finance is to be used for expansion, with Sowho saying there is a great opportunity for AI in Africa as it enables the development of digital infrastructure in lieu of the physical structures that many countries are challenged to deploy.
“There is an opportunity to use AI to overcome these logistical and infrastructure challenges, achieve scale across the diverse populace,” she said.
The potential impact of AI in Africa, then, is massive. Yet many of these companies are utilising these technologies in relatively undisrupted markets, and often with users that are less than tech savvy. Do they take some convincing of its merits and ease of use?
From a DataProphet perspective, Cronje said it depends on whether customers are collecting data and whether their processes support the additional information that AI can find in that data.
“We find that some organisations will have the necessary infrastructure, data and the right resources to start using AI effectively. In addition to that, they have problem statements that speak to AI and an immediate need,” he said.
Beyond manufacturing, however, there are challenges. Paterson said farmers live in the “tangible world”, perhaps more so than any other industry.
“This can create some initial hurdles when bringing our products to a farmer that is used to feeling, seeing and smelling everything on the farm,” he said.
Paterson’s own background in farming helps in the conversion process, and Aerobotics has designed its products from the ground up with the farmer in mind.
“Additionally, while building our products, we consult our customers and industry-leading agronomists to be sure that we are solving real world challenges,” he said.
In the end, it comes down to the product in question and the extent of the problem it is solving, says Sowho.
“Users will respond to a right product. Adoption is determined by the product-market fit and the value proposition to the end user,” she said. “The tech does play a role in adoption but only as an enabler.”