An [agent](/guides/building-custom-agents/) on its own is something you run by hand. An **automation** is an agent wired to a **trigger** so it runs by itself when something happens. You configure automations from an agent's **Automations** tab — each one pairs a trigger with the agent and the output it should produce. This guide covers the three trigger types and how to run and manage an automation.

## The three trigger types

When you add a trigger, you choose one of three kinds:

- **Schedule** — *"Fire on a recurring time, like every Monday at 09:00."*
- **Event** — *"Fire when something happens — like a document landing in a folder."*
- **Email** — *"Fire when an email arrives at a per-workflow address."*

A single automation uses either a schedule *or* event/email triggers — schedule-based and event-based triggers aren't mixed in one automation.

### Schedule

A schedule fires the agent on a cadence. You can set a **one-off** date and time, or a **recurring** cadence by choosing an interval and the days of the week it should run, with a configurable timezone. More elaborate cadences are expressed as a cron schedule that the agent authors for you — you describe what you want in chat and it sets the expression.

### Event

An event trigger fires when something changes in your documents. The available events are **Document — Uploaded** (*"When a new document is uploaded."*) and **Document — Updated** (*"When an existing document is updated."*). You can optionally scope the trigger to a specific folder, so it only fires for documents landing there.

### Email

An email trigger gives the automation its own private inbound address — of the form `agent-…@inbound.regnora.com` — generated when you first save it. Anything sent to that address runs the agent on the message. You can optionally restrict which senders are allowed, so only mail from addresses you trust will fire it.

**Manage:** [Agents](https://app.regnora.com/agents)

## Running and managing an automation

Each automation has an **enable/disable** toggle, so you can pause one without deleting it. You can also **Run now** to fire it by hand — useful for testing, or for an event-style automation you want to run against a chosen document on demand.

Every automation keeps a **run history**: each run records whether it succeeded or failed, what triggered it, and what the agent produced, and you can replay a run to inspect or re-do it. That history is the first place to look when an automation didn't do what you expected.

## Seeing scheduled runs

Scheduled automation runs — and any calendar entries an agent publishes — show up on the agent's calendar, which you can subscribe to from your own calendar app. See [Calendar feeds](/guides/calendar-feeds/).