Skip to main content
Everything in TorchRunner that runs on its own has a name: a Runner. A status report that assembles itself every Friday, a watch on a renewal window, a chase on the open items of a close. A Runner is a pre-built piece of work that knows its own trigger, its own steps, and what done looks like. You direct it and correct it, the same way you would any work in TorchRunner.

​
The hierarchy: Persona, Runner, Step

Runners live in a three-tier structure that mirrors how you would staff the work with people.

Persona

Every Runner is owned by one of the team. The Controller owns the close chasers, the Marketer owns the outreach follow-ups. The owner is on the row, so you always know who to hold accountable.

Runner

The unit you actually manage: direct it, pause it, mute it, correct its output. One Runner, one job.

Step

Inside a Runner sits its procedure: the trigger, the conditions, the actions. Steps are visible and editable. A Runner is not a black box.

​
One Runner, two lenses

The same Runner can be read two ways, and both are the same object underneath, consistent with one object, many lenses. The canvas lens shows the Runner as a flow: trigger into conditions into actions. This is the lens for building and debugging the logic. The prose lens reads the same Runner as a written procedure, the way you would document a process for a junior. This is the lens for reviewing the standard and catching what the flow view hides. Edit in either lens. There is no second copy to drift.

​
What starts a Runner

Two trigger families.

Event triggers

A record is created, a stage changes, a field updates, a note lands, a task completes. The Runner fires off the change, immediately.

Scheduled triggers

A date approaches, a deadline passes, a record goes quiet, a task goes overdue. The Runner fires off the clock, on a sweep.

​
Status that tells the truth

The quiet failure is the expensive one: an automation that stopped working three weeks ago and nobody noticed until the client did. Every Runner carries a live health state on its row, derived from its actual recent runs.
StateMeaning
HealthyRecent runs completed clean.
WarningA failure in the last few runs. Worth a look.
FailingThe most recent run failed. Needs you.
PausedYou switched it off. It will not fire.
MutedSilenced for a window you chose.
Never runConfigured but has not fired yet.

​
Accountable automation

Every run is recorded: when it fired, what it did, how it ended. You can pause a Runner, resume it, or mute it for a window when a client situation makes the automation temporarily wrong.
Automation in TorchRunner is earned, not assumed, and Runners is where that principle is enforced. A Runner graduates to less oversight only after its run history has proven it safe and boring on your practice. See direction and correction.