About 12 000 Shoprite Group employees are using generative AI tools, and the retailer has built its own platform to route their requests to whichever model can handle them most effectively and cheaply.
Chris Shortt, group chief technology officer at Africa’s largest retailer, told TechCentral’s Meet the CIO podcast that the group runs AI in production rather than in pilots, and does so in two distinct areas: customer-facing applications, and internal work, which is itself split between general employee productivity and core retail decision-making.
On the core retail side, the aim is to shorten the lead time between insight and action in category management, pricing and promotions, and in replenishment. Shortt said that work is in production and used in decision-making daily.
Shortt joined Shoprite in January 2024 as group CIO, a title since changed to CTO – a change he described as a naming convention covering the same accountability. He spent more than 12 years at UCS Solutions and was group chief information and technology officer at Pick n Pay before moving across.
For employees, the group built its own platform. Shoprite AI presents staff with an interface Shortt likens to ChatGPT, while gateways behind it pick a model based on the complexity and scope of the request.
The reason is cost. “If you don’t watch the economics of these things they can really get out of hand very fast,” Shortt said, describing a least-cost computing approach to selecting which model answers a given query. Roughly 12 000 users are on the group’s generative AI tools.
Shoprite employs more than 174 400 people, so the 12 000 represents a fraction of the total workforce, which is mainly employed in its retail stores.
No constraints
Shortt has put no constraints on how his software teams use AI, and said the group is actively encouraging it. The gains are not confined to writing code: he said the whole software development lifecycle has compressed, starting with how requirements and business analysis are defined before anything reaches an engineer.
The detail he offered about how his engineers work is unusual for a listed company executive. They deliberately play models off against one another – working from a specification with Claude Code, then cross-checking the result against Gemini, for example.
“Between those two, you kind of get a very elegant outcome, because there’ll always be something that each model will find on each other,” he said, adding that the different large language models each have “a slight personality” that shows through.
That has changed the shape of a delivery team. “Generally now, when we are doing a piece of work on a product feature, we’ll have one or two humans in the mix, and then they’ll have a number of agents that are building with them,” Shortt said. These are “probabilistic models, not deterministic finalities”, he said, and validating what comes out of them remains the responsibility of the person using them.
It is the talent question that Shortt has taken furthest. Shoprite runs a tech academy that sponsors university students from their second year, recruiting into software engineering, data science and industrial engineering. His advice to anyone starting out is to build a skills base that puts them “in a position to tell the machines what you need from them” – and, crucially, to validate what the machines produce.
Then he goes a step further. Shortt said he recently told the group’s chief people officer that Shoprite may have to plan for a future in which new recruits arrive with their own agents.
“How do we enable some of those agents in future? And do we do things like part of the remuneration package as a token allowance, or something like that?” he said. Intellectual property questions would follow, he added. “It’s not beyond the realms of possibility that that can be how people work in the future.”
There is no policy behind this – Shortt was careful to frame it as internal thinking rather than anything decided. But it is a question being asked out loud at South Africa’s largest private-sector employer.
The full Meet the CIO interview with Chris Shortt will be published on TechCentral later this week. — © 2026 NewsCentral Media
