Reliability and operations
The outage that hurts a small business is not the one that throws errors. It is the morning brief that stopped arriving three weeks ago and nobody noticed, because nothing went wrong loudly.
Absence is treated as a signal.
The platform alarms on the brief that did not arrive and on a scheduled sender that has gone quiet, not only on error counters. It runs a sweep every fifteen minutes that builds incidents out of signals that never threw an exception, so degradation surfaces before it becomes an outage.
Separately, synthetic journeys run continuously against the deployed platform, including authenticated paths, so a broken sign-in is discovered by us rather than reported by you.
It surfaces with an action attached.
An operations console, not log tailing
There is a separate internal application for watching this: fleet views, per-customer detail, investigation, economics, platform health, and release health. Failure modes appear there with something you can do about them, rather than only in a log file.
Traces you can follow
Agent runs emit traces that can be filtered down to a single customer and a single run, so a question about what happened on Tuesday has an answer.
Costs are watched too
A cost-spike detector, a drift detector, a quota watcher and a billing reconciler run continuously, so an unusual bill is noticed by the platform before it is noticed by you.
Failed runs diagnose themselves
A failure produces a plain-language diagnosis, and a schedule failing the same way repeatedly pauses itself rather than burning your budget on a timer.
There is no published uptime figure on this site, because publishing one we have not measured properly would be exactly the kind of number this company refuses to invent. The alarms above are real and you can ask us what they have caught.
What does running it actually cost?
Reliable and unaffordable is not a good outcome either. Here is how the spending is kept in hand.
What it costs to runAll areas