O-OPS
OperationsThe context an agent has to read before it is allowed to decide anything.
O-OPS holds the asset hierarchy, the work history, the parts, the as-builts and the certificates, and it is where a decision ends up as a closed job. Without it, an agent is reasoning about equipment it knows nothing about.
O-OPS holds the asset register, the work history and the evidence, and is where a decision stops being a recommendation and becomes a closed job you can prove.
A finding without a record is just an opinion.
Knowing a blade has leading edge erosion tells you very little on its own. Knowing it was repaired eighteen months ago, that the repair used a particular coating, that the same coating is failing on three other turbines, and that the warranty runs out in April, tells you what to do. That second kind of knowledge lives in a register, and most portfolios keep it in a shared drive.
One record, four states
The same work order the other modules keep handing off to, followed from the moment it is raised to the certificate that closes it.
- 01
It is raised with its origin attached
Where it came from, which asset it is against, and what already happened to that asset. A technician opening this does not have to go and find out whether it has been touched before.
o-opsIt is raised with its origin attached1/4 - 02
It is planned around reality
A window when the block can actually come down, a crew that is free, and the part confirmed on the shelf. Scheduling work that cannot be done is how backlogs are built.
o-opsIt is planned around reality2/4 - 03
It is executed and evidenced
The checklist is the method, ticked in the field on a phone that does not need signal. The photos are attached as it goes, not reconstructed from someone's camera roll a week later.
o-opsIt is executed and evidenced3/4 - 04
It closes as something you can show
A certificate, a time to repair, and a trail that survives an audit. This is also the row that makes the next prediction better.
o-opsIt closes as something you can show4/4
From asset to closed job
Six things the register has to hold for a decision to end as work that actually happened.
Asset register and hierarchy
Site down to serialised component, with the parent and child relationships that let a finding roll up and a cost roll down. The naming set at design is the naming used in year ten.
Work orders
Raise, assign, schedule, dispatch and close, with the evidence attached at the point of work rather than typed up afterwards.
Preventive and predictive scheduling
Calendar, runtime and condition based, so a job can be triggered by a date, by hours run, or by O-ARC seeing drift.
Parts and inventory
What was fitted, what is on the shelf, what it cost, and what the lead time is. A work order that cannot be resourced is a work order that will slip.
Field execution
Digital checklists and ITPs on a phone, offline first, with photo capture that attaches to the component rather than to a camera roll.
Document control and compliance
As-builts, certificates, test evidence and inspection reports under revision control, with the audit trail a regulator or an insurer expects to see.
The work order, end to end
Most of what goes wrong in maintenance goes wrong in the handoffs. Each of these states is a place a job can stall, so each of them is a place the record shows you it has.
Raised
From a finding, a schedule, a threshold or a person. Whichever it was is recorded.
Triaged
Severity confirmed, priority set against everything else open on the portfolio.
Planned
Parts checked, access arranged, crew and window assigned.
Dispatched
On somebody's phone, with the history and the evidence they need attached.
Executed
Checklist completed in the field, photos and readings captured at the asset.
Closed
Verified, costed, and written back so the next decision knows it happened.
What it lets you actually measure
These are the numbers an owner is asked for and usually has to assemble by hand. Here they fall out of the record.
MTTR
Mean time to repair, measured from raised to closed rather than from when somebody remembered to open a ticket.
MTBF
Mean time between failures, per component class, so a bad batch is visible as a pattern rather than as bad luck.
PM compliance
Preventive work completed inside its window, as a proportion of what was due.
Backlog age
How long open work has been open, weighted by severity, which is the number that predicts next year's failures.
- 01O-EYEDetectInspection & digital twin
- 02O-ARCAnalyseSensors, IoT & telemetry
- 03O-OPSOperationsThe system of record
- 04actsO-AGENTDecide & actThe agentic layer
Every action writes back to the record, so the model sharpens with every job.
Questions
We already have a CMMS. Do we have to replace it?
No, and we would usually advise against it. Where an EAM or CMMS is your system of record, O-OPS reads from it and writes back through its API. The register still needs to exist somewhere; it does not need to move.
Does the field app work without signal?
Yes. Checklists, history and photo capture are offline first, and sync when the crew gets back into coverage.
How do as-builts stay current?
They are a by-product of capture rather than a document somebody maintains. When O-EYE flies the site, the built state updates against the same asset model.
Can we export for an audit?
The audit trail, the evidence and the document revisions export together, which is the form an auditor asks for rather than a database dump.