No AI Without Information Architecture
Every organisation wants the top of the ladder. Almost none want to pay for the rungs. This series is about the work underneath an AI programme — collection, governance, semantics, ownership — and, just as practically, how you get a public-sector or enterprise client to engage with that work at all, when leading with the architecture is the reliable way to lose the room. Drawn from field experience across South African public sector and enterprise delivery.
10 articles
- Part 1The photograph of a spreadsheetSenior managers photograph an Excel sheet, paste it into a slide, and email it upward. That is the monthly report. The licences to do better were bought years ago.6 min read
- Part 2Start with one formWe used to sell big systems to people who had no idea what we were talking about. Now we ask for one paper form. It is a smaller sale to the same destination — and it actually closes.6 min read
- Part 3Show the art of the possible before the architectureReal client data takes months and approvals nobody has started. Synthetic data shaped to their described pain takes an afternoon — and it moves the meeting off technology and onto their operation.7 min read
- Part 4Sell to the pain, not to the org chartThe technology office is rarely the buyer and frequently the gate. Start there and you get a polite technical conversation with no budget attached to it.6 min read
- Part 5No AI without information architectureEvery organisation wants the top rung. The rung that decides whether the answers are trustworthy is the second one, and almost nobody wants to pay for it.7 min read
- Part 6The semantic layer is your hallucination controlAn agent asked for revenue returns a number either way. Whether it matches what the finance director means by revenue depends on whether a definition exists somewhere the agent can read.7 min read
- Part 7Most of the data you collect is never usedYou pay for the storage, the pipelines, and the field time of everyone filling in the forms. Collection without a decision attached to it is not a data strategy — it is an expensive filing habit.6 min read
- Part 8Where the data is allowed to liveFor a growing number of South African public bodies, the hosting question is settled before the architecture starts — and it removes options rather than reshuffling them.7 min read
- Part 9Architecture decides the billSelf-service works. That is the problem. Four analysts running scheduled reports becomes four hundred people asking whatever they like, and the cost scales with curiosity.6 min read
- Part 10From one form to a data strategyNine articles, one motion. The sales sequence and the architectural dependency order turn out to be the same order — which is why they do not have to be traded off against each other.7 min read
Working through any of this in your own organisation?
These articles come out of delivery work rather than theory. If the problems in this series look familiar, a short scoping conversation is usually the quickest way to work out what is worth doing first.
Talk to CloudNala