Skip to main content
An automation is an Internal Agent task that runs on a schedule, like every weekday at 9am, or a trigger, like a GitHub PR merging (see Tasks). Common use cases: daily performance reports, knowledge gap detection, learning from tickets a human handled, proposing knowledge updates when a PR merges, following up with customers after a bug is fixed, and outage detection.
The Automations page
Set up automations on the “Automations” page. It shows all existing automations and includes templates for common setups. When an automation finds a high-leverage improvement, the Internal Agent creates a todo with the relevant context and attributes, assigned to the right user or team - see Improve.

​
Trigger

An automation fires based on one of these trigger types:
  • Schedule: runs on a fixed cadence, such as every weekday at 9am. Good for recurring reports.
  • Events:
    • GitHub event: runs when an issue is created or a pull request is merged in a connected repo. Good for release-impact checks or syncing technical changes into the Customer Agent.
    • Linear event: runs when an issue is created or moves to a specific status. Good for closing the loop with customers when a bug is marked Done.
    • Notion event: runs when a page is created or updated in a connected workspace. Good for keeping the Customer Agent in sync with policy or runbook changes.

​
Prompt

The prompt is the instruction the Internal Agent runs every time the automation fires. For example:
Every weekday at 9am, look at yesterday’s transferred tickets, identify patterns the Customer Agent should have known, draft memory updates for review, and post a summary in #customersuccess in Slack.

​
Auto-approve

By default, every action in an automation run waits for manual approval. You can set specific tools to auto-approve in an automation: on the Automations page, edit the automation via ”…” and pick the actions to allow under “Auto-approve.” Tasks started by the automation inherit its auto-approve settings.

​
Visibility

An automation is either private (visible to you) or shared (visible to your team). Triggered runs inherit the automation’s visibility.

​
Run limits

You can cap how many times an automation is allowed to run before it stops automatically. Useful for short-lived monitoring, such as “for the next 3 days, check yesterday’s tickets and DM me if the new refund policy looks off.” When the limit is reached, the automation auto-disables.