On this page
A mining services contractor, a drilling company or a metals business runs on crews, equipment and site rules, and the office holds it all together. Rosters set who flies in for which swing. A training matrix records which tickets, inductions and competencies each worker holds for each site. Timesheets come back from remote sites on paper or in an app and have to agree with the shift reports, the roster and the client’s contract before anyone is paid or anything is billed. Operations administrators, payroll, procurement and accounts each rekey part of it. Rostering, maintenance and ERP software already do much of the core work for many firms, and where they do we build around them. Automation takes the checking, chasing, matching and record-keeping around that software and runs it as fixed, tested software. AI is used only where something has to be read, such as a handwritten timesheet or a supplier’s certificate, and its output is checked before anything is paid, billed or recorded against a worker.
These are patterns we see in mining and metals businesses, not client case studies. Your process gets its own map in the assessment.
Rosters, swings and site access
A fly-in fly-out or drive-in drive-out roster looks settled until someone is sick, a rig moves or the client changes the shift pattern. The operations administrator rebuilds the crew list in a spreadsheet, rebooks flights and rooms, and sends the site its access request for each new worker, often the night before. Rostering software already handles the roster itself for many contractors, and where it does we feed it rather than replace it. Most of the work around a crew change follows rules. When the roster changes, fixed software builds the travel request and the site access request for each person on the swing, and checks each worker against the training matrix so a worker without a current ticket or induction for that site is flagged before they travel. AI reads client and travel provider emails for changed dates, shifts and bookings, and drafts the update. The operations administrator approves the roster before anything is sent, and a clash, such as a worker due on two sites, goes to the site supervisor to decide.
Timesheets, shift reports and client billing
On a drill site or a mine services contract, the same shift turns up on three documents: the worker’s timesheet, the daily shift report (a drill plod, for a drilling crew) with metres, consumables, standby and breakdown time, and the roster. Payroll needs the timesheet to agree with the roster. The client needs an invoice that follows the contract’s schedule of rates and agrees with the shift reports their supervisor signed. When these come in as photos of paper sheets from a remote site, someone spends days matching them up. AI reads the handwritten timesheets and shift reports into structured records, and the supervisor confirms each shift. Fixed rules then compare the hours across the three documents, apply the contract rates for drilling, standby, mobilisation and consumables, and list approved work that has not yet been billed. Any mismatch goes to payroll and the supervisor before pay or billing runs. Your payroll software’s award or agreement interpretation still does the pay calculation, and we leave that to it.
Crews, tickets and competencies
The training matrix is usually a large spreadsheet: a row for each worker, a column for each ticket, licence, induction and site-specific competency, and a date in each cell. It is maintained by hand, so it is out of date by the time anyone reads it, and an expired ticket is often found at the site gate. Workers upload their cards and certificates to a register, AI reads each one for the type, class, issuer and expiry date, and anything unclear goes to the training coordinator to confirm. From that register the matrix is generated rather than typed. Reminders go out to the worker and the coordinator well before each expiry, and the roster flags any worker due on a site whose requirements they no longer meet. Whether a worker is competent for a task stays with your supervisors and assessors.
Each client site has its own induction, its own medical and fitness-for-work requirements and its own access forms, and they change. The operations administrator keeps a folder per site, sends the same documents again for every new starter, and chases the site for confirmation. Nothing in that needs AI. A register holds each site’s requirements, the access pack for each worker is assembled from the register and the training matrix, and outstanding inductions and confirmations are chased on a schedule keyed to the mobilisation date. Changes to a site’s requirements are entered by a person, and every pack sent is recorded against the worker.
A new driller’s offsider, fitter or operator needs a contract, a TFN declaration, a super choice form, bank details, emergency contacts, copies of tickets and whatever pre-employment checks the client requires, often within a few days of being offered the job. HR or onboarding software covers much of the collection, and what it doesn’t reach is checklists and dates. A checklist confirms every record is in before the first swing, missing items are chased by text, and the worker’s tickets go straight into the training register. There is no judgement in that work, so there is no AI in it. Your onboarding officer deals with anything still missing.
Equipment, parts and purchasing
Rigs, light vehicles, compressors and mobile plant each have a prestart check and a service interval measured in hours or kilometres. Prestarts are done on paper or in an app, defects are written in a notebook or mentioned on the phone, and hour meter readings reach the workshop late. Where you already run maintenance software, we feed it rather than duplicate it. Fixed software collects the prestart and meter readings, raises a work order for each reported defect, and flags plant approaching its service interval to the workshop. AI reads free-text defect notes and photo captions into the work order, and the workshop lead checks it. Whether a machine is fit to operate stays with your qualified people on site.
Drill bits, rods, muds and additives, wear parts and safety gear are ordered constantly, often by a supervisor calling the office from site. The office raises the order, chases the supplier, arranges freight to a remote location and tries to tie the cost back to the right contract. Raising and tracking orders follows rules, so it runs as fixed software. Requests come in through a simple form, purchase orders go to your approved suppliers at agreed prices, each order is chased until it is confirmed and dispatched, and the cost is coded to the contract or rig. AI reads supplier replies for confirmed dates, substitutions and price changes. The procurement officer approves orders above the limits you set, and any substitution goes to the supervisor before it is accepted.
Suppliers and freight carriers invoice in their own formats, sometimes weeks after delivery, and an accounts officer matches each invoice to the purchase order and the delivery docket before it is paid. Freight surcharges, part deliveries and price changes all turn into emails back to site. Capturing the invoice is well served by packaged tools. The matching after capture is rule work. Each invoice is checked against the order and the goods received within your tolerances, approvals are routed to the person who raised the order, and the cost is posted to the right contract. AI reads only the invoices that capture tools cannot.
In a metals business, steel, alloy and other material usually arrives with a mill or test certificate, and customers expect each certificate to follow the material through cutting, fabrication or processing to their delivery. The certificates arrive as PDFs and scans, and someone files them and matches heat or batch numbers by hand when a customer asks. AI reads each certificate for the grade, standard, heat or batch number and supplier, and the quality officer checks the reading before it is recorded. Fixed software then links the certificate to the receipt, the job and the dispatch, so the right certificates go out with each delivery. Whether material meets the specification is a decision for your quality people.
Safety, reporting and the office
On a remote site, an incident or hazard is reported to the supervisor, written on a site form and sometimes entered into the client’s system as well. Your own safety team may get it days later, and corrective actions are tracked in a spreadsheet. Workers and supervisors get a form on their phone that works offline, and each report is routed to your safety coordinator and the client contact you nominate, with follow-up dates set and overdue actions escalated. AI summarises long free-text reports and suggests a category, which the safety coordinator confirms. The investigation, any notification to a regulator and what changes on site stay with your safety coordinator and site management.
Mining clients usually want a monthly report on each contract: hours and metres or tonnes, standby and downtime, safety statistics, training status and open actions. Someone builds it by hand from the shift reports, the training matrix and the incident register. Because those records already exist, the extraction, calculations and layout run as fixed software on a set day each month. AI drafts a few lines on what changed for the contract manager to edit. The figures come from your systems, not from AI, and the contract manager approves the report before it goes to the client.
The operations inbox holds crew change requests, client access confirmations, travel bookings, supplier dispatch notices, timesheet photos and invoices, all mixed together. Someone reads each message and forwards it, and a missed message can mean a worker turns up without site access. AI classifies each email and matches it to the worker, the site, the purchase order or the contract. Fixed rules then route it to operations, payroll, procurement or accounts, or straight into another automated process. Anything the classifier is unsure of goes to a review queue for a person to sort.
Retyping between the roster, payroll, ERP and client portals
A crew change is entered in the roster, again in payroll, again in the travel booking, again in the client’s access portal and again in the job costing system. A timesheet is typed into payroll and its hours are typed again onto the client invoice. A change made in one place rarely reaches the others, so a worker removed from a swing still has a flight booked. Each hop looks small, so nobody measures it as one process. Almost all of it is fixed software, with field mappings between systems, a scheduled sync, reconciliation checks and an exception queue for records that do not line up. Where a client portal has no way to exchange data, the entry stays with a person, and we prepare it so it can be checked and entered quickly. AI is used only to map free-text notes into the right field. We measure the whole chain as one process first, so the cost is visible before anything is built.
If one of these costs you real money every month, describe it in four answers.
Thinking about what a process actually costs you?
The assessment is small, fixed-price, and tells you the real number, whether or not you ever build.
Worth a 30-minute conversation.