WeBuyCars, South Africa’s leading vehicle trading platform, is embracing artificial intelligence (AI) and machine learning to revolutionize how cars are bought and sold.
“We’ve got two big AI drivers in WeBuyCars, called blue and orange,” Wynand Beukes, chief digital officer at WeBuyCars, says.
“Orange is the customer facing Large Language Model (LLM), and that’s on our website, also with agentic AI functionality for internal use. Blue is a collection of machine learning models that contains all the propriety information and pricing models of WeBuyCars, where we take a lot of factors into account like our historic
data, purchasing and selling history, market analytics and trends. All those factors we use in pricing models, we update on a weekly basis.”
Since late 2022, AI has become a mainstream technology as OpenAI’s ChatGPT rose to global prominence with applications across various industries and sectors of the economy.
Technology continues to advance rapidly, with AI agents generating significant buzz in the industry.
An AI agent is a software system that uses artificial intelligence to act autonomously and proactively on behalf of a user to achieve specific goals. Unlike simple chatbots or reactive programs, an AI agent can perceive its environment, reason and plan its own course of action, use external tools and APIs, and learn and adapt over time
without constant human intervention.
For WeBuyCars, the AI agents are already proving their worth.
According to Beukes, Blue “has bought just over 2,800 cars autonomously without any human pricing involved. We’re scaling that up as we go.”
AI agents, also known as agentic AI, are good at breaking down a complex task into smaller steps, deciding on the most efficient way to accomplish them, and even selfcorrecting its own output, making it a highly capable and independent tool for automating complex workflows and solving problems.
The adoption of agentic technology speaks to the digital transformation WeBuyCars has experienced over its 24-year history.
Founder brothers Dirk and Faan van der Walt spent years physically moving from one part of South Africa to another, finding vehicles, negotiating prices, finding buyers and selling them on. 17 years into the enterprise saw Beukes’ entry as technology chief, looking to turn essentially manual processes into a fully fledge enterprise system that could manage all major aspects of the business centrally.
Describing the situation in 2018, he says “everything was managed on a Google sheet” from inventory management to debtors and creditors.
“And they sold over 2,000 vehicles this way a month. The problem was it had reached breaking point. We couldn’t just throw more people at the problem. We knew we had to digitalise this big company. One of the first tough decisions that came along was, do we go out and buy a traditional ERP [enterprise resource planning] system so that we can get control of the stock. 2,000 cars is a lot of cars to get control of and it’s difficult to do that on a Google sheet. Or do we build our own software?”
The answer to this question came in the form of a fully customised system, built from the ground up that is used to manage all aspects of the business and operations.
Now, WeBuyCars offers a seamless car-buying experience by blending e-commerce with a physical presence. It leverages data and technology to run its 17 “vehicle supermarkets” and just under 100 “buying pods” nationwide.
“When we started, there was nothing. We didn’t have to deal with legacy, which was one of the great things,” says Beukes.
“We wanted to be in control of the software and the second thing is we wanted to be in control of the data”.
When measuring the success of this technology investment, the mindset is simple: “We’re never satisfied.”
Beukes describes their process of coming up with and testing new features quickly.
“When we get a new idea, we want to test it quickly using our experimentation platform. This allows us to expose the idea to a small segment of the market or employees, gather rapid feedback, and learn fast. If it fails, it fails fast — with minimal cost.”
The approach is similar when it comes to the use of AI.
A major internal partnership exists between WeBuyCars’ technology and its marketing division. The company has become a digital marketing powerhouse, leveraging its website, search and social media as a means to generate buyer and seller leads.
Once the leads are in, the technology can take some of the burden off human hands, who are then more able to deal with complex cases.
“If we can automate a certain percentage of our lead or buy lead management, that’s
where a customer wants to sell a vehicle to us…we can, with the same number of people, handle so many more leads,” Beukes says.
“And then we let the petrolheads, the humans, focus on the fringe cases.”
For example, “There’s no reason why a human has to price a Polo Vivo, because the volumes are so high. But in fringe cases like a 1974 Mercedes Benz, that’s a difficult thing to price. It needs understanding. So that’s where the human factor will always stay involved.”
