Over the past few years I have seen a great many knowledge management projects. Almost all had the same goal: collect knowledge, file it, make it findable. And almost all ended at the same point. The wiki is full, the drive is tidy, and the most important question of the week still ends with one single person.
That is not a failure of the tools. It is a misunderstanding about what actually needs securing.
What knowledge management promised
The idea was good and still is: what an organisation knows should not be lost. It turned into filing structures, wikis, document management, search systems. Those who introduced them meant well, and for contracts, policies and manuals it works.
The problem is the rest. According to Panopto and McKinsey, 42 percent of job-related knowledge exists only in individual heads. And that rest is not random. It is the exceptions, the warning signs, the reasons why something is done the way it is. Exactly what makes an organisation able to act.
What intelligence management does differently
Intelligence management starts with a different question. Not: where do we file this? But: can this organisation still decide well tomorrow if its most experienced person leaves today?
Three differences follow. First, expertise is captured actively, in conversation with the people who hold it, instead of waiting for someone to write it down. Second, it is verified and given a named owner and an expiry date, so it is clear what applies. Third, it is taken to where decisions are made: to people as training or answers, and to AI systems as a reliable base.
Why it is becoming urgent now
Two developments turn an old problem into an acute one. The experienced generations are retiring across Europe, taking with them knowledge that was never documented. And organisations are betting on AI that would need exactly this knowledge. According to MIT NANDA (2025), 95 percent of AI pilots show no measurable return. In my observation that rarely lies with the model. It lies with the AI not knowing the organisation’s context.
AI cannot know anything nobody has told it.
What this means for you
You do not have to switch off your wiki. The question is a different one: what do you measure success by? If the answer is the number of documents, you are doing knowledge management. If it is how often a critical question still ends with one single person, you have arrived at intelligence management.
Why wikis and SharePoint alone are not enough for this is covered in the articles on the Confluence alternative and SharePoint. What it means for leadership when coordination becomes cheap and judgement expensive is described in the article on coordination and judgement.
What remains
I am convinced that in five years we will ask organisations a different question from today: how well they decide when the answer is in no document.
How well they manage their knowledge, nobody will ask any more.
