Wagering, payments and data platforms fail on the schedule that hurts most: overnight, on a public holiday, in the last minutes before a Saturday metro meeting. The question is not whether someone is monitoring. It is whether the person who answers can read the code, has the access to act, and knows what the system is supposed to be doing.
The Operations Desk is staffed by our own engineers. It is not a subcontracted first line reading a script and creating a ticket for somebody else.
What the desk covers
We run the platforms we build, and we take on platforms we did not build. The second case is the harder one and we treat it as an engagement in its own right: read the system, find out what is actually monitored, write the runbooks that were never written, and produce an honest list of what will break before it does.
Cloud-native, hybrid and legacy systems are all in scope. Legacy usually means undocumented rather than old, and documenting it is part of the job.
Coverage
The desk runs continuously. Cover spans Australian racing hours, including Saturday metro meetings and overnight processing windows, and shift handovers are written rather than verbal.
Response targets are set per engagement against your severity definitions. We would rather agree them with you than publish a number here that turns out not to apply to your system.
How we work inside your process
The desk fits your existing architecture, ticketing and change governance rather than requiring a parallel one. Escalation paths, approval requirements and out-of-hours authority are agreed during onboarding. During an incident, nobody should be working out who is allowed to authorise a rollback.
Incidents produce changes
Every incident gets a record, a severity and an owner. Recurring incidents are defects, and they go into the backlog of whoever owns that service with the evidence attached. An operations desk that absorbs the same failure every fortnight without fixing it is a cost, not a service.
When this is the wrong call
If your system has no monitoring and no runbooks, buying overnight cover first is the wrong order of work. The desk cannot see what is not instrumented. In that case the engagement starts with an instrumentation and runbook piece, usually alongside Platform to Platform, and cover starts once there is something to watch.