A Foundational Capability Almost No One Has Built
Most organisations think their knowledge lives in their systems. It usually doesn’t.
What lives there is the visible layer: the files, the dashboards, the tools, the folders, the process docs, the data. Useful, often essential, but not the same thing as knowledge.
The actual knowledge, the choices, the reasoning, the judgement that holds the business together, lives somewhere else. In heads, in conversations, in DMs. Sometimes… in the heads of people who’ve already moved on.
Not what's written. What's understood.
The trade-offs that were weighed and rejected. The exceptions that quietly became rules. The judgement that tells someone what to do when the playbook runs out. The context that keeps the next decision coherent.
Without it, decisions drift off-strategy. Calls get made without the framework, the steer, or the grounding to support them. The tangents compound. Sometimes nobody notices until the drift is structural.
Which is fine, until it isn't.
Because organisational knowledge is not the data or information a business stores. It's the living accumulation of its choices, the meaning behind them, and the shared reasoning that helps it make the next one. Artefacts like SOPs and process docs are simply the current representation of those accumulated choices.
And the choices are what differentiate the business. Not the systems. Not the information. Choices made, and choices the organisation will still make.
Which means the most valuable thing the organisation holds is also, almost always, the most fragile.
I've watched this play out across teams over the past decade. The specifics differ, sector, size, stage, but the shape doesn't.
And the world is changing faster; teams are more distributed, roles turn over faster, decisions get made in more places by more people across more tools. The "organisational memory" that used to travel through proximity (overheard context, quick clarifications, knowing who to ask) doesn't travel the same way anymore.
So we compensate. We create local versions of truth to keep moving. We optimise for speed in the moment, not durability over time. We capture outcomes, but lose the rationale, the trade-offs, the assumptions. We retreat into private docs, inboxes, and DMs because shared knowledge starts to feel less trustworthy.
None of this is malicious. It's just how humans adapt under cognitive load.
But the result compounds: not a lack of information, but a lack of shared, trusted understanding. Execution gets uneven. Teams depend on key individuals. Decisions slow as conversations get repeated, just to recover context.
Now layer AI on top. Connect it to the drives, the wikis, the chat, and watch what happens.
It surfaces the contradictions. The outdated processes. The parallel versions of truth. The drafts that were never authoritative. The decisions made without context. The policies no one really owns.
The technology works. The knowledge environment doesn't. AI doesn't reduce organisational complexity. It amplifies whatever structure (or lack of structure) is already there.
Better models will keep coming but the differentiator over the next few years won't be access to AI, most organisations will have similar access, most will end up using the same agents and platforms. It will be whether the AI has anything coherent to reason from.
Some teams have started to notice this. I've seen posts and had conversations recently of teams "codifying how the business works" into markdown files and AI prompt libraries.
It's a good instinct. But it tends to scale into one of two failure modes. Either it stays with solo operators who become very capable on their own, while the team around them doesn't develop with them. Or it becomes a shared location no one governs and no one fully trusts, the same problem the organisation had before, just in a new file format. The instinct is right. The architecture is missing.
Even more concerning: as the team evolves, will the team itself have anything coherent to reason from?
Which raises a deeper question. Maybe the most important one a leadership team can ask right now: Do we have a shared, governed understanding of what we know? Something the team can actually reason from, to make the next choice coherently.
What is authoritative? What is evolving? What is contextual? Who owns it? How does it improve over time? What survives a few key resignations?
The past decade put data on the agenda. Most organisations now have data architects, data governance, data strategy, disciplines that exist because we recognised numbers and structured information needed deliberate design.
This decade’s problem is not data. It is the layer above. It is everything else the organisation knows, and will continue to learn. And, how the organisation develops the foundational capability to steward and govern its knowledge over time.
Knowledge architecture, in other words.
Comments