Artificial intelligence meets industry in South Africa
When the machine learning renaissance started in 2013 – thanks to the development of new graphic processing units accelerating data processing by over 100 times – Frans Cronje and Daniel Schwartzkopff started positioning themselves to fill the gap in the South African market.
“Daniel and I met while doing our undergrad studies at the University of Cape Town and collectively came up with the idea to start an artificial intelligence (AI) company,” says Frans Cronje, the managing director of DataProphet.
“At the time there was hardly any competition in the field and skills were scarce, as you could not even study machine learning here.”
They initially offered machine learning consultancy services to retail, manufacturing and finance companies and from there catapulted into the products market when they started DataProphet in 2015.
“We were relatively new to the game. Working as consultants helped us gain valuable experience and insights into the common challenges faced by these industries, which in turn allowed us to create service solutions for these challenges,” says Cronje.
Another advantage of starting as a consultancy, was that it did not involve large overheads.
By the time DataProphet needed capital to move into the product space, the market was already familiar with the company’s vision and innovations, resulting in the start-up being largely sponsored by existing clients.
“Client companies bought equity in DataProphet, which freed up Daniel and me to worry less about the financial side and focus on the innovation side of the business,” says Cronje.
DataProphet supplies an AI program to improve efficiencies in manufacturing plants.
It collects data from all manufacturing measurables and then uses machine-learning models to predict anomalies, quality errors and areas that can be changed to improve operations.
“The process usually starts with getting all data to a central server, as data is often siloed in different parts of an organisation and not shared between divisions,” Cronje explains.
One of their solutions, for example, helped an engine block manufacturing company to halve its scrap rate in the first month of deployment and reduced it to zero within three months, representing a monthly cost saving of R10m, according to him.
AI solutions are also supplied for service and insurance companies.
“We use machine learning to build profiles of clients and to predict their actions. For the insurance industry, for example, this might help to identify clients who are most likely to default on payments, or the best finance products to offer clients after they have bought specific products,” Cronje explains.
“We have even developed an emotional recognition program for a gaming company to enhance gamers’ gaming experiences.”
DataProphet easily attracted clients after its launch, since machine learning was such a niche field at the time.
“When we started out, there were only two other companies offering these types of services in the country, and I think there are about four dedicated companies today. Each of us specialises in specific fields, so we seldom compete directly with one another.”
Work done for clients led to referrals to other companies. “Local companies, for example, would refer their international divisions to us.
This, in addition to the fact that we are invited to many international industry conferences, has helped us expand our services into the European and American markets,” Cronje says.
The company started out with the three founding members and now has 30 full-time employees. “Our staff is one of our greatest assets. It is a vibrant mixture of people with backgrounds ranging from statistics, to engineering and computer sciences.
Most of them are self-taught in the field of machine learning,” Cronje says. “It really helped that we started out with a strong core team that has always been passionate about the business.”
He identifies a lack of understanding as one of the biggest challenges in the industry: “We have been fortunate that there has been a lot of interest in what we are doing, yet almost half our time is spent educating people about what AI and machine learning can and cannot do.
People struggle to understand the value that machine learning can create, and that it creates ongoing solutions that might differ one day from another. This is especially a problem if the company is resistant to change.”
What surprised Cronje most about the industry was the warm way in which South Africans are accepted internationally.
“As a South African company, it can be daunting to operate in international markets alongside the likes of IBM or Microsoft. However, I found that we can, and have, outperformed international companies when competing on bids. The level of expertise in South Africa in terms of AI, is very competitive on an international level.”
The company plans to grow its footprint in Europe and the US over the next five years and eventually become a large international player.
“We expect to achieve this goal by supplying bleeding edge AI solutions to support the implementation of the Industry 4.0 [the fourth industrial revolution] to manufacturing hubs across the world.”