Permit-to-work agents for mining
The short version: the permit queue at shift start is where a mine site loses its first hour. A permit-to-work agent pre-checks applications for completeness, validates them against your isolation and hazard rules, flags conflicts with other live permits — and puts only the exceptions in front of the authorising officer.
Last updated:
The problem, precisely
Permit-to-work exists because the alternative is people dying. Nobody argues with the control; everybody argues with the queue.
Applications arrive incomplete — a missing isolation reference, an SWMS revision that’s out of date, a gas test that hasn’t been logged. The authorising officer discovers each gap one at a time, at 6am, with a line of contractors waiting and a schedule already slipping. Meanwhile two permits that conflict with each other — the same isolation point, overlapping hot work and confined-space entry — get caught by vigilance, or don’t.
The control is sound. The administration around it is the bottleneck, and the bottleneck is exactly where corners get pressure.
What the agent does
Pre-check on submission. Every application is validated the moment it arrives: fields complete, isolation references real and available, attachments present, certifications of the named workers current, SWMS revision correct. Incomplete applications go back to the applicant immediately — the night before, not at the gate.
Conflict detection. The application is checked against every live and pending permit: shared isolation points, incompatible work types in the same area, simultaneous operations your rules prohibit. Conflicts are surfaced with both permits side by side.
Queue for authorisation. The authorising officer sees a queue of complete, conflict-checked applications with exceptions flagged — and spends judgement on judgement, not on chasing missing fields.
The officer signs. Always. The agent never authorises a permit. It removes the administration around a decision that stays entirely human — and every check it ran is logged with the permit as evidence.
Permit-to-work today vs with an agent
| Today | With an agent | |
|---|---|---|
| Incomplete application found | At authorisation, 6am | At submission, the night before |
| Isolation conflict | Caught by vigilance | Checked against every live permit |
| Contractor certification | Manual lookup | Validated automatically |
| Authorising officer’s time | Chasing gaps | Exercising judgement |
| Evidence trail | Paper and memory | Every check logged with the permit |
| Shift-start queue | The first hour, lost | Exceptions only |
The line we will not cross
The agent administers the permit process; it does not assess hazards, approve work or touch isolation systems. Anything ambiguous escalates. Any vendor offering to “automate permit approval” is selling you a fatality inquiry — walk away, including from us if we ever describe it that way.
What it connects to
Permit and HSE platforms: existing electronic permit systems where deployed, or a structured layer above a paper process where not.
Isolation registers: to validate references and detect shared points.
Contractor management: certifications, inductions and site access status — the same layer described in vendor compliance.
CMMS: so permits link to the work orders that require them.
Frequently asked questions
Does this replace our authorising officers?
No, and it must not. It removes the clerical load around their decision so the decision gets their full attention. Authority and accountability do not move.
Can it handle our site-specific rules?
That’s the point of a custom build. Your isolation philosophy, your SIMOPS matrix, your permit types and your trigger conditions are encoded as hard rules — and rules are inspectable, so your HSE team can audit exactly what’s being checked.
What about regulator acceptance?
The agent strengthens the evidence trail: every completeness check, conflict check and certification lookup is logged against the permit. A qualified person still authorises. Bring your regulator’s guidance to the audit and we’ll build to it explicitly.
What happens when the system is down?
Your existing paper process remains the fallback, documented and rehearsed. An agent that becomes a single point of failure for permitting is a build error.
How long to deploy?
Typically 6–8 weeks including shadow mode, longer if permits are currently paper-only. The audit tells you which in week one.