Every operation has one: the metallurgist, the maintenance planner, the senior geo who has been on site longer than the plant. When they go, they take a working model of the whole operation with them.
Mining runs on long careers. The people who commissioned the plant, weathered the last three downturns, and know why the crusher behaves badly in the wet season are, in many operations, the same generation, and that generation is heading toward retirement together. That is what makes it a cliff rather than ordinary turnover: the departures cluster, and the replacements arrive with a decade less exposure to the orebody, the equipment, and the site's particular history.
The pattern is made worse by how the industry now staffs itself. Rosters rotate, contractors churn, and the graduate who might once have spent fifteen years absorbing a site's habits now moves on in three. The traditional transfer mechanism, long overlap between generations on the same site, has quietly disappeared, while the knowledge it used to carry has not become any less necessary.
The succession plan usually covers the org chart. It names who steps up, and it schedules a handover period. What it rarely covers is the substance of what is being handed over, because most of it has never been written down anywhere.
The dangerous myth is that the knowledge lives in the documents. Some of it does: the procedures, the drawings, the geological models, the maintenance history. But the part that keeps the operation running is judgement, and judgement looks like this:
None of this is in the document management system. It is in the heads of a small number of senior people, and it is retrievable only by asking them, which stops working the day they leave.
The standard response is a handover period and a knowledge-transfer document. Both fail the same way: they ask a person to summarise a career from memory, in a few weeks, without knowing which questions the future will ask. The expert writes down what seems important, which is the general stuff. The valuable material, the specific judgement calls, only surfaces when a real situation demands it, and by then the person is gone.
Pairing a successor with the expert for six months is better, but it transfers only what happens to come up in those six months. The failure that occurs every four years, the negotiation that happens once a contract cycle, the flood response, none of it may occur during the handover window.
We covered the general version of this problem in how to capture senior staff knowledge before retirement. On a mine site the stakes are simply larger, because the systems are larger and the consequences of relearning by trial and error are measured in downtime.
The approach we build for mining and resources operations works differently. Instead of asking the expert to write a summary, you interview them, repeatedly, in structured sessions built around real scenarios: walk me through the last major failure, walk me through how you plan a shutdown, what would you check first if recovery dropped tomorrow. Recorded conversation captures judgement in a way a template never does.
Those sessions, along with the reports, emails, and records the expert has accumulated, become a private knowledge succession layer: searchable, questionable, and organised, so the engineer facing a problem at 2am can ask how the site handled it before and get the answer with its context and caveats.
The architecture matters as much as the method. Geological data, reserve information, production numbers, and operational history are commercial IP of the most sensitive kind. A knowledge system built on a cloud AI service sends that IP to someone else's servers. The systems we build are self-hosted: the models and the knowledge layer run on hardware you own, inside your own network, and nothing leaves the site unless you decide it does.
The practical question every operations manager asks next is what this costs the expert in time, because the expert is already the busiest person on site. The honest answer: a few hours a week of structured conversation over a period of months, scheduled around the roster, plus access to the documents they already have. The heavy work, transcribing, organising, cross-referencing against the site's records, and building the searchable layer, happens off their desk. The expert talks; the system does the filing.
Just as important is what happens after they leave. A capture project that produces a static archive decays the day it is finished. Built properly, the knowledge layer keeps growing: new incidents, new fixes, and new lessons get added by the people still on site, and the retiring expert's judgement becomes the foundation of the record rather than its entire contents.
Not every departure justifies a capture project. If the role is genuinely procedural, if the knowledge is already well documented, or if the person's expertise is portable industry knowledge rather than site-specific judgement, a standard handover is enough. The same is true if the expert is leaving in three weeks: real capture takes months, and starting too late produces a thin result that gives false comfort.
The economics work when the knowledge is site-specific, expensive to relearn, and held by few people. If losing this person would mean re-running experiments the operation already paid for once, the case makes itself. If it would mean a slightly slower quarter, it may not.
The uncomfortable truth about knowledge capture is that it is only possible while the person is still there and still willing. Every month of delay narrows what can be captured, and the departure dates are usually known well in advance. If your operation has senior people inside five years of retirement, the time to look at this is now, while the option still exists.
The way we start is deliberately small: a paid discovery engagement, fixed fee from AUD $6,000, that identifies who holds the judgement your operation cannot afford to lose, what it would take to capture it, and what the build would cost. You keep the written diagnosis either way, and if you proceed to a build within 30 days, 70% of the discovery fee is credited against it.
The first conversation is thirty minutes: who holds the judgement, when they leave, and whether Wild Systems is the right answer.
Book a 30-minute diagnosis or email info@wildsystems.com.au