Emergency Rescue

Most emergencies are not one bug. They are one bug plus four things nobody knew about the system, and a team that has been awake too long to see straight.

The first job is not to fix it. The first job is to find out what is true.

What happens

You send access — repository, logs, infrastructure, whatever exists. If there is no documentation, say so; that is normal and it is part of what you’re paying to have established.

Before anything starts you get a timeline: what will be turned around, and by when, given honestly rather than optimistically. A date that gets met matters more here than a date that sounds impressive.

We reproduce the failure. Not a plausible story about the failure, the failure itself. Until it can be triggered on demand, any fix is a guess wearing a lab coat.

Then you get a written finding: what broke, the mechanism, the blast radius, and the shortest safe path back to stable. If the fix is twenty minutes of work, the finding says so and you can hand it to your own team. If it is three weeks and a migration, the finding says that too, and you get to make that decision with the facts in front of you rather than at the end of another sprint.

When it is genuinely on fire

Salvage works read-only by default and proposes changes rather than merging them. Your team ships the fix, because a fix nobody on your side understands is a future incident wearing a bow.

An outage is the one case where that can be the wrong answer, and pretending otherwise would be posturing. So there is a break-glass path, agreed in the engagement before it is needed: written authorisation from a named person on your side, a fixed time window, a log of everything touched, and a written handover of every change when the window closes. The log is part of what you receive.

Outside that window the normal rule holds. You get the fix as a pull request with the reasoning attached, and your team merges it.

Why it’s fixed-fee

Because an hourly rate on an emergency is an incentive to take longer, and because you should be able to say yes to this without a procurement cycle. The number is the number. It covers one incident; if the investigation turns up a second, unrelated failure, you will be told before any time is spent on it. If the assessment turns into a repair engagement, the fee comes off the first invoice.

What you’re actually buying

Certainty. At the end of it you will know what is wrong with your system. That is worth more than a partial fix from someone still guessing.