The Corporate Mycelium - What Fungi Can Teach Us About Organisational Knowledge

I was watching a nature documentary recently when the presenter started talking about mycelium.

For anyone unfamiliar with it, mycelium is the vast network of fungal threads that exists beneath the ground. What we see above the surface, the mushroom, is only the visible part.

Underneath it can sit an extensive network intertwined with soil, roots and the surrounding ecosystem.

Naturally, I couldn’t help making the connection to enterprise knowledge.

I say “naturally”, although my partner might use a different word.

She was trying to concentrate on the documentary while I was providing a running commentary along the lines of:

• “That’s exactly what happens with organisational knowledge.”

• “Look, distributed information networks.”

• “This is basically SharePoint, but with better governance.”

Apparently, repeatedly pausing a nature programme to explain its relevance to enterprise AI doesn’t improve the viewing experience. Who knew?

But once I’d seen the connection, I couldn’t unsee it.

Because organisations have their own version of mycelium.

Most of what keeps a business functioning doesn’t sit neatly inside the systems that appear on an architecture diagram. It travels through informal, largely invisible networks.

• It sits in conversations between colleagues.

• It lives inside emails, Teams messages and personal folders.

• It’s carried around in the heads of experienced employees.

• It appears in unofficial spreadsheets, personal notes and documents with names such as “Final_v7_REALLY_FINAL_Approved2”.

And, much like the fungal networks beneath a forest floor, most of it remains hidden until something disturbs the surface.

What you can see isn’t the whole system

Look at a forest and you’ll see trees, plants, leaves and the occasional squirrel behaving as though it owns the place.

What you won’t see is the network below.

Mycorrhizal fungi form associations with plant roots. Broadly speaking, the fungi can help plants access water and nutrients from the surrounding soil, while receiving carbon produced by the plants in return.

The reality is more complicated than the slightly romantic idea of trees running a woodland internet and sending each other supportive messages. Scientists continue to debate exactly how much plants transfer through shared fungal networks, what causes those movements and whether some popular claims have gone beyond the available evidence.

But the underlying point remains fascinating.

The visible organism can depend heavily on a network that nobody sees.

That’s where the corporate parallel begins.

A company might have an official knowledge base, an intranet, a document management platform and an AI assistant. From the outside, it appears organised.

Underneath, though, there’s usually another network.

It’s the person in Operations who knows the documented process doesn’t work in certain circumstances.

It’s the contact centre agent who keeps a personal list of the answers customers actually need.

It’s the compliance specialist everyone messages because the official policy is impossible to interpret.

It’s the technician who knows that “restart the system” really means restart three particular services in a very particular order while quietly hoping nobody asks why.

This hidden network often keeps the organisation alive.

The problem is that nobody can see it properly.

Your employees are routing around broken knowledge

People are remarkably good at adapting to poor systems.

When official knowledge is incomplete, outdated or hard to find, employees don’t usually stop working. They find another route.

• They ask a colleague.

• They save a copy locally.

• They create a cheat sheet.

• They bookmark an old email.

• They copy a previous answer.

• They develop a workaround and pass it quietly around the team.

That ingenuity is valuable, but it can also disguise the scale of the underlying problem.

Management sees that work is getting done and assumes the knowledge system is working.

It may not be.

• The people are working.

• The informal network is working.

• The official knowledge could be sitting in the corner eating biscuits and contributing very little.

This creates an uncomfortable question:

How much of your organisation’s performance depends on knowledge that your organisation doesn’t officially know it has?

Then AI arrives

Before AI, the gap between official knowledge and the hidden network was already a problem.

AI makes it much more significant.

Connect an AI assistant to the official repositories and it gains access to what has been stored. It doesn’t automatically gain access to everything employees have learned about whether that information is trustworthy.

It may find the approved procedure, but not the workaround everybody uses because the procedure is obsolete.

It may retrieve the latest-looking document, but not know that the real answer sits inside a message written by someone who left two years ago.

It may summarise three sources beautifully without recognising that experienced employees quietly ignore two of them.

This is one of the central misconceptions surrounding enterprise AI.

Businesses often assume that connecting AI to their documents gives it access to their organisational knowledge.

It doesn’t.

It gives AI access to their documents.

Those aren’t necessarily the same thing.

Documents are the visible mushrooms. Organisational knowledge is the whole network underneath.

AI can expose the wrong layer

There’s another complication.

Informal knowledge networks contain valuable information, but they also contain rumour, outdated habits, local interpretations and workarounds that should’ve disappeared years ago.

Not everything travelling through the corporate mycelium is healthy.

One team may have created an effective solution to a recurring problem.

Another may still be following a process introduced during the pandemic because nobody ever officially told them to stop.

A third may rely on a spreadsheet built by Martin in 2019.

Nobody remembers exactly what the spreadsheet does. Martin has left, the macros are protected, and opening it produces the sort of warning usually associated with handling radioactive material.

Yet the business still depends on it every Tuesday.

This means the answer isn’t simply to connect AI to everything.

Doing that may capture more knowledge, but it may also give equal credibility to approved policies, obsolete workarounds, personal opinions and “helpful” documents that should’ve been retired a decade ago.

AI needs access to knowledge, but it also needs context:

• Where did this information come from?

• Who owns it?

• Is it still current?

• Is it approved?

• Does another source contradict it?

• Is it intended for this country, product, customer or situation?

• Is it official guidance or simply something somebody found useful once?

Without that context, greater access can create greater confusion.

What happens when a key connection disappears?

Natural networks can be resilient because they don’t always depend on a single route.

Corporate knowledge networks often appear resilient for the same reason.

But sometimes the resilience is an illusion.

Ask yourself what would happen if the employee everyone relies on suddenly left.

Not in six months after a carefully managed handover. On Friday.

• Could the organisation identify what they knew?

• Could it find the documents they trusted?

• Could it see which official instructions they routinely corrected?

• Could it explain the decisions stored only in their inbox?

• Could another employee take over without learning everything through trial, error and increasingly desperate messages?

Many businesses talk about key-person dependency as though it applies only to senior executives or highly technical specialists.

In reality, it can apply to the person who knows how the returns process actually works.

Or the colleague who understands why two policies contradict each other.

Or the employee who has spent eight years quietly connecting information that the company’s systems keep apart.

Remove that person and the business hasn’t simply lost an employee. It has broken part of its hidden knowledge network.

The problem isn’t a lack of information

Most large organisations aren’t short of information.

They are drowning in it.

The real problem is that the relationships between pieces of information are often invisible.

• A policy exists in one system.

• A process guide exists somewhere else.

• A regional version has been copied into a shared folder.

• A training deck quotes an older edition.

• A chatbot draws from a PDF that nobody realised was still available.

• An experienced employee knows they’re inconsistent, but the systems don’t.

This is where the mycelium analogy becomes particularly useful.

Individual documents matter, but so do the connections between them.

• Which documents refer to the same subject?

• Which one is authoritative?

• Which versions have diverged?

• Which content depends on another piece of content?

• Where is knowledge flowing?

• Where has it stopped?

• Where is the network carrying something harmful?

Traditional document management often treats each file as a separate object.

AI knowledge readiness requires us to understand the network.

Building a healthier corporate knowledge network

This is where index comes in.

index helps organisations see and improve the knowledge network that sits beneath their AI, search, customer service and operational systems.

It begins with Scope, where we understand the knowledge landscape, the business objectives, the systems involved and the areas where poor information creates the greatest risk or cost.

Then Scan analyses content across repositories to surface problems such as:

• Contradictory guidance

• Duplicate and near-duplicate content

• Outdated documents

• Broken links

• Missing ownership

• Unclear terminology

• Poor machine readability

• Content that isn’t suitable or reliable for AI

This gives organisations something they’re often missing: visibility.

Not simply another list of files, but a clearer picture of how their knowledge estate behaves.

Solve then supports structured remediation. That might mean consolidating duplicate content, resolving contradictions, repairing links, assigning ownership, improving machine readability or establishing a canonical source that both people and AI can trust.

And through ongoing monitoring, Sustain helps prevent the knowledge network from quietly deteriorating again after the initial clean-up.

1. Because it will deteriorate.

2. People create new documents.

3. Policies change.

4. Teams reorganise.

5. Products evolve.

6. Employees leave.

7. Systems migrate.

8. Somebody inevitably uploads another file called “final”.

Knowledge governance isn’t a one-off gardening project. It’s an ecosystem that requires continuing attention.

AI is only as healthy as the network beneath it

The current AI conversation focuses heavily on models.

• Which model is fastest?

• Which one has the largest context window?

• Which one produces the most impressive demonstration?

Those questions matter, but they overlook what sits beneath the surface.

An AI system may be extraordinarily capable, but it still depends on the knowledge it can access and the context surrounding that knowledge.

If the underlying network is fragmented, contradictory or decaying, AI won’t magically repair it.

It may simply allow the problems to travel faster.

That’s the real lesson of the corporate mycelium.

The health of the visible system depends on the health of the hidden network below it.

Your chatbot, search engine or AI assistant is the part everyone sees.

Beneath it sits the much larger network of documents, decisions, conversations, workarounds, relationships and accumulated human experience that keeps the organisation functioning.

Understand that network and AI becomes far more useful.

Ignore it and you may spend a great deal of money building a very sophisticated mushroom.

And yes, I did explain that final point during the documentary.

My partner was delighted.

Or at least, that’s how I’ve chosen to interpret the silence.

index helps organisations discover, understand and improve the knowledge beneath their AI.

Through Scope, Scan, Solve and Sustain, we identify unreliable information, resolve the issues that matter and provide continuous oversight to prevent knowledge quality from quietly deteriorating again.

Because better AI doesn’t begin with a bigger model.

It begins with healthier knowledge.

Find out more at www.index-ai.net

Contact us at contact@index-ai.net

by Paul Tucker