How to manage reactive maintenance jobs from logged to closed

A reactive maintenance job runs cleanly from logged to closed when every request is captured in the same form, given a priority from the contract and not from whoever shouted, owned by one person at every stage, and closed only with evidence. Most of the work between those points is chasing: the engineer for a date, the subcontractor for a job sheet, the client for approval of a quote. That chasing is written, repeatable work an AI worker can take on. Emergencies, phone calls, spending decisions and anything about safety stay with your coordinators and contract managers.

Where does a reactive job stall?

A reactive job starts when somebody at a client's site reports a fault: a blocked toilet, a door that will not lock, a spill, a failed light. It arrives by email, by phone, through the client's portal or as a word to a cleaner on site. A helpdesk coordinator logs it in the CAFM or job management system, such as Joblogic or MRI Evolution, and from then on a clock is running, because the contract sets a time to respond and usually a time to fix.

A job rarely stalls on the repair itself. It stalls at the joins between stages, when it is logged without enough detail to send anyone, assigned but never accepted, attended but not reported, finished but still open on the system. Each shows up at month end as a missed response time in the client's KPI pack.

The job, step by step

1. Capture. Site, exact location, what has failed, since when, who reported it and how to reach them, access hours, a photo where there is one. Today the coordinator does this. If any of these is missing the job goes back for questions while its clock runs, so make them mandatory on every form and in every email reply.

2. Prioritise. Each contract has its own priority levels and its own times. Write them out as a table per contract: this kind of fault, at this kind of site, is this priority. Then two coordinators give the same job the same priority.

3. Assign. To your own engineer or a subcontractor, with everything from step one attached. The job is not assigned until somebody has accepted it and given a time.

4. Attend. Somebody watches the response time and chases before it passes, not after. Where a first visit cannot fix the fault, the job needs a reason recorded and a next date, or it will sit.

5. Approve extra work. Where the repair needs parts or a quote, and your contract sets a spending limit for reactive works, anything above it goes to the client for approval. Days are lost here.

6. Close. A job sheet, time on and off site, photos of the finished work, and a message to the person who reported it. No evidence, no close.

Which steps can an AI worker take on?

On a reactive helpdesk an AI worker works with a coordinator who checks its work. It can read each request that arrives by email or portal, match it to the client and site, and reply at once with the questions from step one that were not answered. It can set the priority from your contract table and note on the job why, check whether the same fault is already open, and pass the job to the coordinator's queue complete. That half of the work is described under helpdesk ticket triage.

After assignment it can chase the engineer or subcontractor for an accepted time, warn the coordinator about any job approaching its response time with nobody booked, send a quote to the client for approval and chase the answer, ask for the job sheet and photos after the visit, tell the reporter the job is done and ask whether the fault is fixed. That half is maintenance request coordination. At month end the same record becomes the jobs page of the client's pack, which a contract manager checks and sends, as set out under management reporting.

What waits for a person, and why

• Emergencies. Keep a written list: a smell of gas, flooding, an insecure building, anything involving injury. These go straight to a person, by phone. A smell of gas is reported to the gas emergency number at once, by that person, before anything is logged.

• Phone calls. A request made by phone is taken and logged by your team, and so is any call to an engineer who is not answering messages.

• Who goes. Choosing the engineer or subcontractor is the coordinator's unless you have set a rule for it.

• Money. Quotes, spending above the limit, and who pays when a fault is outside the contract are a contract manager's.

• Safety and disputes. Whether an area is taken out of use, a client who says the fault is not fixed, and any complaint.

What to have ready before you start

You can tighten this job tomorrow without us. Write the capture fields from step one at the top of your helpdesk's email template. Put each contract's priorities and times on one page and pin it where the coordinators sit. Agree the emergency list with your contract managers. Decide what counts as evidence for closing a job, and tell your subcontractors that invoices without it will wait. Then, once a day, list every open job by the time left on its clock.

The other roles a contractor's office can hand over are on our page for facilities management and cleaning, and the case for doing so is in scaling a cleaning or FM business without hiring.

For an FM contractor an AI worker would complete each request, keep every job moving against its response time and gather the evidence to close it, and your people would decide who goes and what is spent. Is your helpdesk spending its day chasing when it should be coordinating? Talk to our agent.