“Every system we looked at just bolted a chatbot onto the same admin. I did not need my coordinator to type faster — I needed to stop hiring another coordinator every time we won a site.”
— WA mechanical services contractor, $19M revenue, 70 workers
From Hierarchy to Intelligence: The AI-Native Operating Model for a Mining Back Office
For 2,000 years, growth meant adding people
Block — the company behind Square and Cash App, with Jack Dorsey and Sequoia's Roelof Botha — recently published the clearest account we have read of a problem every business quietly runs into. For two thousand years, from the Roman army to the modern corporation, growing an organisation meant adding people to carry information up and down a hierarchy. A leader can only manage a handful of people, so more work means more layers, and more layers mean slower information flow. They call it span of control, and it has been the governing constraint of every organisation on earth.
Their argument: for the first time, the coordination a hierarchy exists to do — routing information, pre-computing decisions, keeping everyone aligned — can live in the system instead of in layers of middle management. People move to the edge, where judgement and relationships live. The routing in between gets done by intelligence.
Read the original: Block — From Hierarchy to Intelligence. It is written about a company with thousands of staff. The same shape applies to a mining supplier with thirty.
Your back office is that coordination layer
For a $10M–$50M WA mining supplier, the hierarchy is the back office. The coordinator, the HSE officer, the estimator and the accounts clerk exist to route information between the work and getting it done. Every mobilisation, every audit pack, every timesheet and every invoice passes through one of those desks. That is your span of control — and it is why winning the next site has always meant hiring the next coordinator.
- The coordinator is the one person who can mobilise a FIFO worker correctly — so mobilisation throughput is capped at one desk, with gate rejections at $18,500 each when volume outruns care.
- The HSE officer is the one person who can produce the pack a WorkSafe (LGIRS) inspector or a BHP auditor asks for first — so compliance capacity, not field capability, caps how many Tier 1 sites you hold.
- The estimator is the one person who can assemble a compliant tender — so tenders answered per month are capped by one desk, and an unanswered tender is an unwon contract.
We wrote about this constraint in detail in Why Admin Is the Hidden Ceiling on Mining-Supplier Growth. The operating-model lens adds the next question: if the back office is a coordination layer, what happens when the coordination moves into the system?
Copilot or system — the difference that matters
Most mining software using AI today gives everyone a copilot: a chatbot bolted onto the same process, so your coordinator works it a little faster. It is useful, but it does not change the operating model — information still flows through the same person, so growth still means adding people. AI-native is the other way around: the desk runs in the system itself.
Bolt-on AI (a copilot)
- A chatbot bolted onto the same admin process your coordinator already runs.
- Helps one person type faster, but every job still routes through that one desk.
- Win the next site and you still hire the next coordinator. The ceiling does not move.
AI-native (a system)
- The coordination itself — mobilisation, compliance packs, timesheets, tenders, invoicing — runs in the system.
- Seventeen agents carry the admin across the four desks, 24/7, with no queue.
- Win the next site and the work goes up; the back office does not. The ceiling lifts.
Your people move to the edge
This is not a headcount-cut story. When the routing moves into the system, your people move to the edge — the place where the work actually wins and keeps contracts. On a mine site that edge is concrete: the relationship with the BHP superintendent, the judgement call on a tricky tender, the problem on the tools that no model can see, the trust built face to face in the donga.
The coordinator who spent the day re-keying portals and chasing signatures spends it on the client and the crew instead. The admin churn that drove good people to leave goes to the agents; the work that keeps them stays with the people.
The system keeps one live record
A coordination system is only as good as the record underneath it. DrilLedger keeps one live, timestamped account of who did what and when — actor, time and Award clause on every approval — configured around your Xero, your templates and your rosters. When WorkSafe (LGIRS) pulls the register or a Tier 1 auditor asks for proof, the evidence is already there, not stitched together in the donga at 8pm the night before.
That live record is what lets the coordination run without a person relaying it — the same role a manager used to play, done by the system.
What the shift is worth
Recovered admin labour, full 17-agent suite
$555K–$1.1M+/year — recovered admin labour, not a subscription cost
The upside
You take the next site, contract and 50 workers without adding the next admin hire — same back office, higher margin
The structural difference
A cost advantage competitors cannot match by hiring harder, and growth no longer capped by how fast you can staff the office
The copilot makes your coordinator faster. The system means you do not need the next one.
Frequently Asked Questions
What does AI-native mean for mining software?
AI-native means the coordination work a back office exists to do — routing a mobilisation, assembling a compliance pack, calculating an award timesheet, building a tender, chasing an invoice — runs inside the system itself, not as a chatbot bolted onto a person who still does the job by hand. The difference matters: a bolt-on copilot makes one coordinator faster, so you still hire the next coordinator when you win the next site. An AI-native system carries the admin so the back office stops scaling with the work.
Is this just a copilot for my admin staff?
No. A copilot sits beside your coordinator and helps them work the same process faster. DrilLedger runs the desk itself — the FIFO mobilisation, the WorkSafe (LGIRS) audit pack, the MA000011 timesheet, the tender, the invoice are produced by the agents, not typed by a person with AI assistance. That is the difference between making the bottleneck a little faster and removing it.
Will AI agents replace my office team?
The opposite of how it usually plays out. The admin routing in between — the part that caps how much work you can take on — moves into the system. Your people move to the edge: the site, the client, the tender, the relationships and the judgement calls a model should not make on its own. You stop needing the next admin hire to grow; the people you have do the work that actually wins and keeps contracts.
How is this different from adding ChatGPT to our process?
Adding a general chatbot gives everyone a faster keyboard inside the same hierarchy. It is useful, but it does not change the operating model — information still flows through the same people, so growth still means adding people. DrilLedger is built the other way around: the coordination lives in the system, configured around your Xero, your templates and your rosters, with a timestamped record of who did what and when. The full 17-agent suite recovers $555K–$1.1M+/year of admin labour — recovered admin labour, not a subscription cost.