Triggers

An agent starts on schedule, on demand or from your tools

Schedule in a few clicks, run by hand, or let an application or a coding tool trigger it. Every launch leaves a run to read back.

New schedule

A frequency in plain words, not cron

You pick the agent, the frequency, the time and the time zone. Before you save, the “Next runs” preview shows the exact dates.

  • Every N minutes or hours, every day, some days, every month. The cron expression is still there for those who want it.
  • If the person it runs as leaves the organization, a connection is deleted or the agent is removed, the schedule turns itself off and the list shows the reason.
  • A schedule never uses a personal model key, and an agent that expects input files cannot be scheduled.

Five ways to start, one history

Whatever the trigger, the run is reviewed in the same place, and the history says where it came from.

Started by

  • A scheduleevery Monday at 8 am, every day…
  • The "Run" buttonby hand, from the agent
  • The Chatby asking for it
  • An applicationthrough the API, with a key
  • A coding toolClaude Code, Codex, Cursor
AgentInvoice reminders

Leaves

  • A run to read backsteps, result, cost
  • Files producedreports, drafts, exports
  • An outbound webhookto notify your tools

Names are examples.

From your applications

One API key per use, with an end date

To let an application or a script run your agents, create an API key in the space concerned. It expires when you decide and only opens what you tick.

  • “Expires in” 30, 90 or 180 days, 1 year, or never.
  • “Scopes” to restrict the key, or “Full access” when it is justified.
  • The key is shown only once, when it is created.
  • Claude Code, Codex or Cursor, connected through MCP, can also run your agents.

Outbound webhooks

Your tools know when a run ends

A webhook notifies the address of your choice when a run starts, succeeds, fails, times out or is cancelled. Your tools take over without anyone watching.

  • HTTPS address and deliveries signed with a secret you can rotate.
  • “Send a test” checks delivery before you rely on it.
  • “Deliveries” keeps the history of what was sent and the responses.
  • A webhook notifies, it does not start an agent.

Frequently asked questions

Can an agent start when an email arrives?
Not directly: webhooks notify your tools, they do not start an agent. A schedule can, however, check the mailbox at regular intervals.
Can I run a schedule right now?
There is no button for that on the schedule. You run the agent by hand from its page, with “Run” or “Run with options…”.
What cannot be scheduled?
Agents that expect input files. A schedule cannot use a personal model key either.
What happens to a schedule when its author leaves the company?
It turns itself off and shows the reason. The same goes when a connection it uses is deleted or the agent is removed.

Let's see what it would look like on your side.

Try it on your own in a few minutes, or take thirty minutes with us to scope a first use case.

To try it

Try it for free

Create a cloud account and describe your first agent.

For a guided start

Let's talk about your case

Thirty minutes with us to scope a first use case and its gain.

To audit

Read the code

Apache 2.0. Your technical team can read it all, install it all, keep it all.