When knowledge retires

On the last working day there are flowers, a short speech and applause. What nobody in the room says out loud: almost four decades of experience are leaving the company with that person. Which supplier actually delivers when things get tight. Why line 3 is started up differently than line 4. Who to call at the authority so a permit does not sit around for six weeks. None of it is in a handbook.

This is not an isolated case, it is the normal case for the years ahead. By 2039 around 13.4 million employees will retire. Each of them takes an average of two decades of experience home with them.

What new employees cannot possibly know

The thinking error starts with the replacement. A role is advertised, someone qualified starts, the gap counts as closed. In terms of professional skills that is often true. In terms of experience it never is.

What a colleague learned in twenty years was mostly not learned in a training course but in situations: the complaint that escalated, the order that nearly went wrong, the customer who suddenly agreed on the third call. None of this is written down, because at the moment of learning it did not look like knowledge to anyone. It was simply the working day.

So the new colleague has to walk the same path again. She will take the same detours, ask the same questions and make the same mistakes that were made in the company long ago. The only difference is that the company pays for them a second time.

Lost knowledge costs money, it just never appears on an invoice

Knowledge loss shows up in no cost centre. It is noticeable all the same, in four places:

• Onboarding takes longer. Without documented processes, every new person needs months instead of weeks before working independently.
• Experienced colleagues become an information desk. Constant interruptions cost them time for their own work. It hits your best people hardest.
• Decisions take longer. Once nobody remembers why a process is built the way it is, every change turns into a debate with an open outcome.
• Mistakes repeat. The most expensive mistake is the one the company already made once and then forgot.

Do the maths for a single position: three months of extra onboarding, plus two hours a week that an experienced colleague spends answering questions. With several departures a year that quickly adds up to a full role nobody ever applied for.

Tapping knowledge while there is still time

The decisive factor is timing. Starting four weeks before retirement is damage control. Starting two years before is knowledge transfer. Four levers work together here:

Documentation and preparation. Not as a shared folder, but structured by task. A fifteen minute video about a process is faster to produce and more useful than ten pages of prose nobody reads. What matters is that the recordings are findable afterwards and sit in a sensible order, the way the editor lays them out.

Mentoring and tandems. Experience transfers best through shared work. A tandem over six to twelve months achieves more than any farewell documentation, because the questions come from real situations instead of a checklist.

Technology that does the collecting. People dislike documenting for a hypothetical future. A system that captures the answer at the moment the question is asked gathers knowledge without a separate project.

A culture where sharing is normal. As long as knowledge is seen as personal job security, nobody will hand it over voluntarily. That is not a question of tooling but of incentives.

What AI can already take over

The biggest practical step forward is not better text documents. It is that knowledge becomes retrievable and collectable without anyone having to type.

An AI agent can chase project status, for example. Instead of a project manager calling five people one by one, the agent calls or messages regularly, asks for the current state, follows up on anything unclear and records the result. What emerges on the side is a chronicle of the project that still explains in two years why a given decision was made.

The reverse direction works the same way. If a question comes from sales that really belongs to engineering, an AI assistant finds the right place in the existing material and answers directly. Only when nothing is there does the question go to the experienced colleague, and her answer then moves into the knowledge base. Next time nobody has to ask at all. How that plays out with your own content is shown on the page about AI support.

The effect matters more than the technology behind it: knowledge is collected exactly when it is needed anyway. Nobody has to sit down and reflect on what they know.

Knowledge transfer does not happen on its own

The most common mistake is the instruction "please write down what you learned in forty years". That task cannot be completed. People cannot recall what they know unconsciously, and a blank page makes it worse.

What works is the specific question. Not "what do you know?", but "how do you proceed when a customer asks for a custom build?". Not "document your area", but "show me the next case and we will record it".

On top of that come two barriers no tool solves on its own:

1. Time. Knowledge transfer that is supposed to run on top of daily business does not happen. It needs fixed hours in the calendar, just like a customer meeting.
2. Recognition. Whoever passes on knowledge gives up part of what makes them indispensable. If that is not seen and rewarded, holding back is a perfectly rational response.

Both are a leadership task. Without it, every system stays empty, however good it is technically.

Where to start

You do not have to start with forty years of company knowledge. Three steps are enough to get going:

1. List the people leaving in the next five years. That is one report from HR and takes an hour.
2. Ask each of them for the three tasks nobody else can do. No more than three. That list is your order of priority.
3. Record the first task the next time it comes up anyway. No extra appointment, no preparation.

After four weeks this secures more than any documentation project that stalls in its kick-off meeting.

Everyone who will retire by 2039 is in the company today. That is the good news: the moment to secure their knowledge has not passed. It is simply not unlimited.

If you want to see what this looks like for your processes, book a demo. We will go through which knowledge is at risk in your company first.