Search interest in legal AI South Africa is rising because the work day is full of short statutory questions. Overtime under the Basic Conditions of Employment Act. Processing conditions under the Protection of Personal Information Act. A clause that needs to match the Act that actually applies. The risk is not that models write badly. The risk is that they write fluently from the wrong corpus.
South African legislation sits next to UK, US and other texts in the same open web. A generic answer can borrow a familiar phrase from another system and still sound local. If the product does not keep South Africa in its own corpus, you cannot tell whether the sentence came from the statute you meant.
NOMOS Legal AI is built for that constraint. Pick South Africa. Ask the question. Read the portions returned from licensed national, provincial or municipal legislation. Open the citation. If the text does not cover the point, the honest outcome is a named gap, not a filled one.
We are not publishing thresholds, POPIA condition lists, or overtime formulae here. Those live in the Acts in force, and they move with amendments and facts. The durable skill for lawyers is the workflow: retrieve first, verify the excerpt, then draft.
If you are evaluating AI for lawyers in South Africa, ask one screening question: can I open the exact provision the answer used? If not, you are reading a summary, not research. More on the workspace is on Product and Trust.
This is research assistance, not legal advice, and not a substitute for reading the full provision in force.