When South Africa's AI practitioners choose a tool for their own use, they pick Claude ahead of ChatGPT. That is the finding of DayOne's first State of AI Delivery ZA 26 report, released on 9 November, which surveyed 282 people who build and use AI in South African businesses. 68% said they use Anthropic's Claude themselves, against 46% for OpenAI's ChatGPT.
At work, the order is the other way round: 83% of their organisations use ChatGPT, and 58% use Claude.
Insights on Claude vs ChatGPT
Put side by side, the gap is large. ChatGPT drops 37 points between company use and personal use, while Claude rises 10 points, giving it a 22-point lead when practitioners choose for themselves. Google Gemini is used by 46% of organisations. Respondents could pick more than one tool, so the figures add up to more than 100%. AI is now part of the daily routine for this group: 87% use AI tools several times a day.
"I basically can't see a world in which I don't use AI every single day for my daily tasks and my job," one respondent said. The report suggests many practitioners choose their own tools, sometimes alongside what their company provides and sometimes instead of it. Data from report sponsor Velocity Labs helps explain how that happens. 75.5% of respondents said more than half the people in their organisation use AI daily, yet 40.1% said nobody in particular is responsible for AI, and only 7.4% have a dedicated AI lead.
DayOne also found that 84% of AI budgets go on tooling and licences, while only 3% of underused tools are blamed on speed or accuracy, which it reads as organisations buying software that people route around. The report does not say why practitioners prefer Claude. The people surveyed are a specialist group: 62% are in technical roles, 36% work in engineering, 66% work at companies of 50 people or fewer, and 67% are based in Cape Town or Johannesburg. They were surveyed between 2 and 16 October.
What others are saying about Claude vs ChatGPT
DayOne, an AI build studio and network of AI professionals founded by Liesel Bester, summarises its overall finding as "adoption is solved, delivery isn't", arguing that the gap is people and resources rather than technology. The report was made possible by OfferZen, Pepperplane, Velocity Labs, Altron AI Factory and Digiata. These results sit alongside a very different picture of everyday users: in our look at what South Africans ask ChatGPT, OpenAI's own data showed how the wider public uses its tool.
The company buys one tool, the team uses another
The striking part is not which tool wins, but the gap between what companies pay for and what their most skilled people actually choose. When most organisations have nobody in charge of AI, that gap is easy to miss. It costs money through licences that go unused, and it creates risk, because company information can end up in tools the business never approved, which brings in the obligations we set out in our guide to POPIA and AI at work.
The practical fix is simple: before renewing AI licences, ask your people which tools they actually use and why, and give one person clear responsibility for AI. That is also the thread running through Monday's look at small business AI.
You might also like our piece on SA not using ChatGPT for search, AI agent swarms in Minecraft, and SA's AI policy.
Get more SA tech and business news and subscribe to The Open Letter.


