Generative AI has changed what’s possible in software development, but how a team uses it matters far more than whether they use it at all.

OpenAI has been fired

We started out with ChatGPT like everyone else. Not any more. OpenAI has been fired and Anthropic is all around our office: on every screen, in every editor and in most of our conversations about how a piece of work should be done.

Anthropic’s models, every day

We use Anthropic’s Claude models on a daily basis. They help us write and review code, investigate bugs, draft documentation, analyse data and build dashboards. AI is not a side experiment at CodeGuys. It is part of how every project gets delivered.

Domain knowledge is crucial

AI will happily produce software that works. Whether that software can be maintained is another matter. Without a computer science background to guide it, the implementation will be working but unmaintainable: no sound architecture, no thought given to security, data integrity or scale, and nobody who understands why it was built the way it was. That is a disaster waiting to happen in a production environment.

This is where experience counts. Our developers know what to ask for, how to judge what comes back and when to reject it. The AI does the typing. We remain responsible for the engineering.

What it costs us

As of October 2026 we are burning AI tokens at a rate of R6 000,00 per month. That covers subscriptions, token-based purchases and other vision and sound models. We consider it money well spent, because our clients get better software, delivered faster.