
In an industrial SME, the retirement of a key employee is both the most predictable and the most poorly prepared event in company life. Predictable, because the date is known six months, a year, sometimes two years in advance. Poorly prepared, because everyone postpones the handover until the farewell drinks.
This article details what is really lost when an expert retires, why the transfer window is shorter than you think, and how to organize the capture of their know-how without asking them to write a single line.
The knee-jerk reaction is to think about the job description: tasks, procedures, tools. But all of this is generally documentable and often already documented. What leaves with a key employee is everything else:
This tacit knowledge represents, according to knowledge management studies, the majority of an expert's skill capital. And it is precisely what no job description contains.
On paper, a retirement announced a year in advance leaves plenty of time to get organized. In practice, three mechanisms reduce this window to a few useful weeks:
Most companies organize a file handover: a few meetings between the departing employee and their successor, a summary document, a site tour. This is necessary, but highly insufficient, for a simple reason: the successor doesn't yet know what questions to ask. The real questions arise three months, six months, or a year after the departure, when facing a concrete situation that the handover had not anticipated. And at that point, there is no one left to answer.
This is the structural flaw of all synchronous methods: they assume that all useful knowledge can be transmitted during the shared presence window. Experience proves otherwise.
Skillsay reverses this logic. Rather than transferring live to a single person, knowledge is captured in a database searchable by everyone, forever.
Specifically, Olivia, the AI voice interviewer, conducts a series of short sessions with the future retiree, spread over their final months. They have nothing to write or prepare: they answer orally, and the AI follows up, clarifies ambiguous points, and cross-references their answers with documents already uploaded to the platform. Each session is transcribed, structured, and added to the company's knowledge base.
The day the successor faces their first unknown breakdown, they ask their question in natural language and receive, within seconds, an answer built on their predecessor's experience, complete with sources. The expert has left, but their know-how is still answering.
An announced retirement is the only major risk of knowledge loss whose date is known in advance. This is an opportunity: it leaves time to organize a serious capture effort, provided it starts several months before the departure and does not rely on writing. A few AI interview sessions are enough to turn decades of experience into an asset queryable by the entire team.
Is a retirement coming up in your company?