Automations: schedules, triggers & one-off tasks

Run an agent once by hand, or put it on a schedule or trigger so it shows up and does the work on its own.

Last updated July 14, 2026

An agent can work two ways: you can run it once, right now, for a job you hand it, or you can automate it so it shows up on its own. Both use the same engine and produce the same kind of result. The only thing that changes is what starts the run. This page covers all three ways to run an agent: a one-off task you start by hand, a schedule that runs on a clock, and a trigger that fires when something happens.

Three ways to run an agent

Every run of an agent starts in one of three ways. A one-off task is simply an automation you start yourself, which is why the output is identical no matter which path you use.

Way to runStarts becauseGood for
One-off taskYou start itTesting, ad-hoc work, trying a change
ScheduleA recurring time you pick arrivesMorning briefings, weekly reports, nightly syncs
TriggerAn event happens in a connected toolNew email arrives, a form is submitted, a webhook fires

You are not limited to one. An agent can have many automations running side by side, and each can be turned on or off independently, so you can pause one without touching the rest. A support agent might run a daily digest on a schedule and reply the moment a new email lands.

Running an agent once

Not everything needs to be automated. Sometimes you just want the agent to do one thing right now and hand you the result, with no schedule or trigger to set up. A one-off run is useful when you are:

  • Testing an agent — confirming it does the job well before you automate it.
  • Doing ad-hoc work — a report or lookup you only need this once.
  • Trying a change — checking how new instructions, Tools, or Knowledge affect the output.
Open the agent

Go to the agent you want to run.

Describe the task

Tell it what you need in plain language, the same way you would in chat — for example, "Pull last month's invoices and total them by client."

Let it work

The agent uses its Tools, Skills, Knowledge, and Memory to complete the job and take any actions across your connected apps.

Review the result

The output is saved and the run is logged, so you can revisit and compare it later.

Schedules vs. triggers

Automations fall into one of two camps. Schedules run on a clock. Triggers run in response to something happening. You can use both on the same agent.

  • Schedules run at a recurring time you pick — hourly, daily, weekly, or on your own cadence. They are ideal for anything that should happen on a predictable rhythm, like a morning briefing or a nightly sync.
  • Triggers run when an event happens in a connected tool — a new email arrives, a form is submitted, or a webhook fires. They let the agent react in the moment instead of waiting for the next scheduled time.

Where to set up automations

Open your agent and go to the Automations tab. From there you can add a schedule, add a trigger, or both. Because each automation is independent, you can switch any one of them on or off without affecting the others.

What happens when a run fires

Every run behaves the same way, whether you started it by hand or it fired on its own. The agent uses its Tools, Skills, Knowledge, and Memory to do the work, and the results are saved for you to review afterward.

The run starts

You start a one-off task, a schedule reaches its time, or a trigger's event happens.

The agent works

It follows its instructions and takes action across your connected tools.

Results land in Deliveries

Any output the agent produces is saved so you can find it later.

The run is logged

Successful runs show in Activity. Anything that fails shows in Errors so you can fix it.

Because one-off tasks, schedules, and triggers all flow through the same history, it is easy to compare a manual test against later automated runs.

Before you automate

Make sure the agent already does the job well when you run it once by hand. Automations repeat whatever the agent does, so a clean one-off run is the best proof that it is ready. While you are at it, confirm the tools it needs are connected and that any Skills or Knowledge it relies on are in place. Once it does the job cleanly, put it on a schedule or a trigger to run the same way on its own.

Runs use credits. Automated and one-off runs both use credits, and there is no extra cost to running by hand versus on a schedule. Keep an eye on frequency for high-volume triggers so runs stay predictable.

Next steps

Was this page helpful?