Kleinhans Digital POPIA-aware implementation
I treat data flow as part of the build, not a policy note after the fact. The work looks at what is collected, where it is processed, who can access it, and whether a local model is the more appropriate path.
About Jason
A hands-on AI builder, not a distant adviser.
I run Kleinhans Digital from Johannesburg for organisations that want practical AI adoption without losing control of client data, internal knowledge or approval flow.
Working principle
I review the risk, advise on the boundaries, and build the automation only when the data path makes sense.
How I work
I bring the operating model I use myself.
My own AI operation combines cloud reasoning, local models, a private knowledge layer and deterministic workflows. That mix matters because not every task belongs in the same tool. Some work benefits from public cloud models. Some work should stay on a controlled machine. Some work should not touch AI at all.
I help South African teams make those calls in plain language, then turn the approved decisions into usable assistants, document flows, websites and automation. I am not a lawyer. I help teams prepare better questions, document practical boundaries and avoid careless handling of personal information.
Credibility
Specific, reviewable, and built close to the work.
I do not sell AI as a vague productivity promise. I start with what your people are already doing, what data they touch, and where the information travels. From there I can advise on a safer adoption path and build the system that supports it.
Local models where they fit
Local AI is useful when the organisation needs more control over prompts, documents, logs and outputs. It is not magic, but it can reduce unnecessary exposure when designed properly.
Builds with judgement attached
I can review, advise and implement. That keeps the recommendations grounded in what can actually run inside a small South African organisation.