The workforce problem came first. When staff work across client sites rather than in one building, the business loses sight of three things at once: whether someone actually turned up, how the hours worked compared to the hours planned and whether those hours have been paid and billed. Most of it gets reconstructed afterwards from memory and paper.

Attendance is handled at the source. Staff start and stop their shift from the app and every sign in records a location ping. The timesheet shows that ping on a map, green where it falls inside the venue geofence and red where it does not. It flags rather than blocks, deliberately. A hard block strands someone with poor GPS or a boundary drawn slightly tight and they still did the work. Marking it instead keeps the record honest and leaves the judgement with a person who knows the site. Actual hours accumulate against planned hours in the same view, timesheets build themselves, and the export to CSV ends the retyping into payroll. Shifts carry payment status and invoiced flags, so the gap between work done and money collected is visible rather than discovered at the end of the month.

Rostering needed two different views because two different questions get asked of it. The weekly planning view is for allocating work, grouped by venue and employee, showing who is on what day and whether it has been paid. The monthly roster view is for coverage across the period. Unavailability comes from the staff themselves through the app rather than through messages that get missed, and surfaces against the calendar when shifts are being created. There is also a dedicated unavailability calendar which switches shifts for unavailability for each employee, facilitating labour planning as required.

The lubrication module is the second half and it has its own constraint: a factory floor does not run on software. Every guide generates a QR code when it is written and that code never changes, so a printed copy can be fixed to the machine permanently. If a single lubrication point fails, the whole report is marked unsatisfactory rather than averaged out, so it lands in front of an administrator. Oils and greases carry an identifying colour that prints on the guide, matched to however the site colour codes its own guns and drums, because the wrong product in a bearing is the expensive failure.

The final phase opened the platform to Chainlink’s clients. Each site operates on its own subdomain with its own invited administrators, seeing only its own staff, guides and reports.