Every established business has them. The person in accounts who knows why one customer is always invoiced differently. The workshop manager who can tell from a drawing whether a job will cause trouble. The office manager who remembers what was agreed with a supplier in 2011 and why. They have been there for decades, and much of what they know has never been written down.
When one of them announces their retirement, owners often realise how much of the business runs on that knowledge. The usual response is to ask them to "write everything down" in their last few months. That rarely works. Here is a more practical approach.
Why "write it all down" doesn't work
Experienced people don't know what they know. Much of their expertise is judgement: noticing that something looks wrong, remembering an exception, knowing who to call. Ask them to write a manual and you get the obvious steps, not the things that actually matter.
There is also a time problem. The people with the most knowledge are usually the busiest, and a long document competes with their day job. And even when a good manual does get written, it tends to sit in a shared folder that nobody opens, because finding the right page takes longer than asking someone.
Capture knowledge the way people actually share it
People explain things well when they are answering a real question about a real piece of work. So build the capture around that.
- Record short conversations, not long documents. Sit with them for half an hour while they talk through a typical job, a tricky customer, or a decision they make every week. Record it, with their agreement. AI transcription tools now make this cheap and accurate.
- Collect the questions people already ask. For a few weeks, note every question colleagues bring to this person. That list is a good map of what the business actually depends on them for.
- Gather what already exists. Old emails, notes in the margin of procedures, job files, quotes they have marked up. These often hold the reasoning behind decisions.
- Ask about exceptions. "When do you not do it the usual way?" is often the most valuable question of all.
Make it easy to find and use
Capturing knowledge is only half the job. It has to be easy to use when someone needs it, usually in the middle of a task with a customer waiting.
This is where AI helps most. A knowledge assistant is a tool that has read your recordings, notes, procedures and past documents, and can answer questions about them in plain language. A newer member of staff can ask "how do we usually handle a late delivery from this supplier?" and get an answer that draws on what your experienced person said, with a link to where it came from.
The important points are that it only answers from your own material, shows its sources so people can check, and says when it doesn't know rather than guessing. Set up that way, it becomes a way of asking your most experienced colleague a question, even after they have left.
Involve the person who is leaving
This works best when the retiring member of staff is part of it, not just a source to be mined. Most long-serving people care a great deal about the business and want it to carry on well. Ask them to test the assistant, correct its answers and add what is missing. It is a good way to recognise what they have contributed, and it catches mistakes early.
Be clear about what is being recorded and how it will be used. Agree it with them in advance, keep it to work matters, and make sure the material is stored securely.
Start well before the leaving date
The best time to do this is a year or two before someone retires, not their last month. It also doesn't need to wait for a retirement. The same approach helps when someone is off sick, when the business grows and new people join, or when you simply want to stop the same questions landing on the same person every day.
A sensible first step is to pick one person and one area of the business: one set of customers, one product line, one process. Capture it, set up a simple assistant, and see whether people use it. Then widen it.
Where we can help
Our Knowledge Capture and Handover work does exactly this: we help you capture what your experienced people know and turn it into something your team can ask questions of, built on your own documents and kept under your control. If you would rather start with a broader look at where AI could help, an Executive AI Session is a good place to begin.
Knowledge walking out of the door is one of the quieter risks in an established business. It is also one of the more fixable ones.
