The experienced buyer knows your biggest supplier overpromises capacity every Q4.
That judgment took fifteen years to build. When they leave, it leaves with them.
Most organisations know this. Most have not solved it because the framing is wrong.
Knowledge transfer programmes ask people to document their expertise on top of a full workload, with no upside, in the final years of a career that was already rewarding without it.
That is not a knowledge strategy.
Here is what changes with AI.
The buyer who has called that supplier every quarter for twelve years can annotate an AI recommendation. That annotation becomes training data. The model learns the exception.
The next planner inherits a judgment that took years to build — without spending fifteen years building it.
Institutional memory becomes permanent. Transferable to every planner who comes after.
What works in practice is treating knowledge transfer as a role, not a favour. Structured time. Recognised contribution.
It is not “please document your spreadsheets before you leave.”
It is “your expertise will outlast your tenure, and we are investing to make that happen.”
That is a conversation most organisations are not having today.
The incoming generation will not build intuition the same way. They will work alongside systems that already carry patterns.
The risk shifts. Proficiency without depth. The ability to operate the system without understanding what it is doing or why. When the recommendation is wrong, the practitioner who never built judgment from first principles will struggle to catch it.
Pair new joiners with the practitioners encoding their knowledge. Build the judgment layer alongside the AI layer.
The talent cliff is real. AI can capture what falls off it. But only if knowledge transfer is treated as a strategy, not a handover.
Does anyone in your organisation have structured time to encode what they know before they go?