The most commented post I have written recently was about AI and project knowledge. The conversation kept returning to a question I want to address directly: what does it actually mean to have the structure in place before AI becomes useful? Here is the practical version. AI surfaces something — a missing reference, a confidence score on a compliance claim, a suggested compliance summary. The engineer reads it.
Now what? If the underlying data has no structure — if the compliance picture lives across documents, spreadsheets and email threads with no defined relationships between them — the AI has created insight that has nowhere to land. The engineer knows something new but has no clear place to act on it. The search problem is reduced. The action problem is unchanged.
When requirements are the fixed reference point that everything else is built around - compliance claims, evidence, open actions, design assumptions - any AI output maps immediately to a specific node. The engineer can locate it, verify it in context, and update it in a way that propagates through the whole picture and organization, including the people. But this requires a tool that can properly manage the data in this way.
Anybody out there with a AI usage story within certification that you would like to share? Which software barriers have you encountered and which solutions you developed (with or without AI) for the tool handling the data? Reach our to us if you wish to dsicuss such challenges and have a look at the solution we designed for them!