Every job, measured against the plan.
Monitors live jobs, detects delays or operational issues in real time, automates routine actions, and alerts supervisors only when intervention is required. Your supervisors see the handful of jobs that have drifted, instead of refreshing a board all day looking for them.
- Watches
- Active jobs, milestones and SLA clocks
- Works in
- Control Center Agent ( Internal)
- Runs on
- Workorder Management and Dynamic Form Builder
- Setup
- Days, once geofences and targets are set
what it is for
The day drifts long before anyone notices.
The plan is set the night before and the day starts moving away from it within the hour. A crew waits at a gate, a start is never clocked in, a two-hour job is still open at four. None of it is hidden. It is all on the board. But someone has to keep looking to find it, and by the time a supervisor does, the customer has usually rung first.
AI Operations Assistant reads that board continuously. It compares every active job with the start time, milestones and SLA clock you set for it, and raises the ones that have slipped. It watches the job, not the person on it. Where you have approved a status update or a delay notice, it posts that; everything else goes to the supervisor you nominated, with the reason attached.
Jobs off plan
Delays surface early
A job that has not started, or is running past its window, is raised while there is still time to move something.
One named person, in order
Each alert goes to the supervisor who owns that job, then up your escalation order if it is not picked up, never into a shared tray.
Jobs are watched, not people
It compares work against the plan. Arrival checks run inside the job window only, and it produces no ranking of individuals.
watch the plan
It reads the day against what you scheduled.
Everything it raises comes from one comparison: the job in front of it against the plan you published for that job. Nothing is inferred from how long someone else took last month.
Scheduled start, expected duration, the milestones on the work order and the SLA clock from the contract. The plan is the one your dispatchers published, so the agent and the board never disagree about what was meant to happen.
A clock-in inside the geofence you set marks the start. Position is read for that check, inside the job window only. There is no continuous location history and no scoring of individual technicians.
Delay reasons, on-hold codes and the fields on your close-out forms are picked up as they are entered, so a crew that has already logged an access problem is not chased about it.
Silence is not treated as a problem on its own. A plant room with no reception, a long drive between sites and a phone left in the van are expected, and the agent waits for the grace period you set.
Plan against the day
work out what matters
Not every slip is worth an alert.
A board that flags everything gets ignored within a week. The agent applies the thresholds you set, works out what a delay puts at risk later in the day, and stays quiet about the rest.
Arrival grace, how long a job may run over, and how late is late are set per job type or per contract. A planned maintenance visit and an after-hours breakdown do not have to share a threshold.
The clock on the contract is tracked as the job runs, so a job is raised while the target can still be met. A breach that has already happened is reported as a fact, not as an alert to act on.
Where a job running long puts the next visit on the same run at risk, both are shown together, so a dispatcher moves one job rather than finding out about three separately.
A job that stays late does not alert again every time it is checked. Once a supervisor has acknowledged it, the agent tracks it quietly and speaks up again only if it crosses the next threshold.
Off plan now
act, then stop
It posts the routine, and hands over the rest.
The agent has a short list of things it may do and no standing permission beyond it. Moving work, committing to a time and anything touching safety are decisions it passes to a person.
On site, on hold, running late, completed. Each one is written to the work order with the time and the reason, so the record reads the same whether a person or the agent posted it.
From your template, to the contact on the job, only for the delay types you have turned on. It does not compose an apology of its own and it does not promise a new arrival time.
The supervisor who owns the job first, then the next person on your list if nothing is acknowledged. Out of hours it follows the on-call order instead of waiting for the morning.
Reassigning a job, splitting a run or pulling a crew off site stays with your dispatchers and operations managers. The agent shows what is at risk and what the options cost in travel; the decision is made by a person.
Approved actions
how it works
Four steps, then it hands on.
The four run continuously on every active job. Only step three acts, and only where you have approved it.
-
monitors
Tracks active work orders against scheduled start times, geofenced mobile clock-ins, and expected completion durations.
-
detects
Identifies operational exceptions such as missed clock-ins, access delays, equipment shortages, or lingering work orders.
-
acts
Performs approved routine actions — updating job statuses, notifying building managers of minor delays, or adjusting follow-up schedules.
-
escalates
Immediately flags SLA risks, safety incidents, or job blockers to operational supervisors via push and dashboard alerts.
where the line sits
What it handles, and what it hands over.
Every active job sits on one side of this line or the other, and the thresholds that decide which are yours.
| Job | What the agent does | Who finishes it |
|---|---|---|
| Start not clocked in | Waits out the grace period, checks the crew is not already on site, then alerts the supervisor who owns it. | Supervisor decides |
| Access delay on site | Records the reason the crew gave and posts your approved delay notice against the job. | Coordinator chases |
| Running past its window | Flags the time over and what it puts at risk on the rest of the run. | Dispatcher reshuffles |
| SLA clock at risk | Raises the job before the target is missed, with the remaining time and the contract target shown. | Operations manager |
| Blocked or unsafe to proceed | Stops and escalates immediately. It posts nothing to the customer and takes no view on the hazard. | Supervisor on call |
| Anything it cannot place | A job with no scheduled time or no milestones is passed on rather than measured against a guess. | Named owner |
you set
Before it runs its first job, three things come from you.
- Geofence radius configurations and arrival grace periods.
- SLA countdown benchmarks and operational milestone definitions.
- Escalation contact matrix defining supervisor notification chains.
None of it is a data migration. It reads the Control Center Agent ( Internal) records you already keep.
stays with your team
The operations assistant tracks telemetry and automates standard alerts. On-site decisions, emergency re-allocations, and staff support remain led by human operations managers.
An escalation arrives with the job, the plan it missed, the reason the crew gave and the time left on the clock, so a manager decides rather than investigates.
rollout
Start it in watch-only mode.
Most teams do not switch every action on at once, and nothing here requires it. The usual order:
Watch first
Connect the board with every action switched off. It tracks starts, delays and SLA clocks and posts nothing, so you read a week of what it would have raised.
Alert the supervisor only
Turn on the internal alert next. Nothing reaches a customer, but the person who owns the job hears about a missed start while it still matters.
Then the approved notices
Release one customer-facing message, usually the delay notice, run it for a fortnight and add the next. Any of them switches back off in one click.
works with
What it runs on, and what it hands to.
read next
Where the day usually slips.
before you switch it on
What teams ask about AI operations assistant.
Only inside the rules you set. Anything outside them - listed under "you set" above - goes to a named person on your team, with the record attached. There is no state where it silently makes a judgement call you have not already approved.
Days, not months. It reads from the records you already keep in Control Center Agent ( Internal), so there is no separate data migration or new system for your team to learn before it can start.
Yes, instantly, and every action it took stays on the record so you can see exactly what it did and reverse it. Turning it off does not remove or hide anything it has already written.
No. It takes the routine, repeatable part of the job off their desk so they spend their time on the exceptions and judgement calls that actually need a person - the work listed under "stays with your team" above.
AI operations assistant runs on your Control Center Agent ( Internal) records, alongside Workorder Management and Dynamic Form Builder, so nothing needs re-keying between systems.
No. It compares jobs with the plan, and the alerts name jobs rather than people. Location is read only to confirm an arrival inside the job window you set, it is not kept as a trail, and the agent produces no league table, no scorecard and no productivity ranking of individuals.
The thresholds are yours, and a job that is running slightly late on a quiet afternoon does not have to raise anything. Each job alerts once, not on every check, and once it has been acknowledged the agent stays quiet unless the job crosses your next threshold.
A gap in signal is treated as a gap in signal. Arrival and form data sync when the phone reconnects, and the agent backdates the timestamps rather than reporting a missed start. Sites you know have no reception can be given a longer grace period so they never raise a false alarm.
See it on your own live jobs.
Book a demo
We will run it against a week of your own jobs, your geofences, your grace periods and your SLA targets.
Book a demoSee the other nine
AI operations assistant is one of ten agents on one job record, from first enquiry to paid invoice.
All AI agentsSee the platform
The field service system underneath: scheduling, work orders, assets, invoicing and the field app.
AI Field Service Management