Joshua Pi'Rwot

Shift handover, quality records and maintenance history beat the chatbot every time. Why cameras on the line are usually the wrong first project in this market.

In a regulated lender the question is never whether the model is accurate. It is whether the decision can be explained. Three uses that pay before scoring does.

Reporting is the obvious AI candidate in the development sector and the riskiest. Where the line sits between aggregate work and beneficiary data, and what to check first.

Switching off the training toggle answers one question and organisations treat it as answering four. What actually happens to your text, and the rule worth remembering.

Four routes to AI capability, priced against published benchmarks. A ten-day scoping study can cost more than a month of embedded engineering and produces a document.

The licence is the small number. Exit cost, verification time and metered charges are the large ones. Ten questions that turn weak answers into contract clauses.

Five failure modes account for most AI pilot deaths in East Africa, and four are decided before the tool is chosen. The largest is verification time nobody counted.

An organisation with no AI policy does not have less AI use. It has AI use it cannot see. The one page that does most of the work, and the four clauses that earn their place.

A national AI policy is in drafting through the Ministry of ICT. Five things worth doing now, and the obligation that already exists and most firms have not resolved.

Published seat prices, a full year-one budget in dollars and shillings, the costs that appear on no quote, and seven ways to cut the bill without pretending you need less.