Workjet
Back to Workjet

Engines (Scheduled Automations)

Engines are cron-style automations that run on a schedule. They execute AI workflows at defined intervals — generate weekly pipeline reports every Monday, scan documentation for accuracy daily, or digest feature requests every morning. Each run is logged with duration, token usage, and status.

Creating an Engine

  1. Click Engines in the sidebar
  2. Click New Engine
  3. Configure the engine:
    • Name: A descriptive name (e.g., "Weekly Pipeline Report")
    • Description: What the automation does
    • System prompt: Instructions for the AI to follow on each run
    • Model: Which AI model to use
    • Schedule: A cron expression defining when to run
    • Timezone: The timezone for schedule interpretation
    • Output channel: Where to deliver results
    • Tools: Which MCP connectors the engine can access
  4. Click Create

Cron Scheduling

Engines use standard cron expressions with timezone support. Common patterns:

Expression Schedule
0 9 * * 1 Every Monday at 9:00 AM
0 8 * * * Every day at 8:00 AM
0 */4 * * * Every 4 hours
30 17 * * 1-5 Weekdays at 5:30 PM
0 0 1 * * First day of every month at midnight

The cron format is: minute hour day-of-month month day-of-week.

Timezone: All cron schedules are interpreted in the timezone you select. For example, 0 9 * * 1 with timezone America/New_York runs at 9:00 AM Eastern every Monday, regardless of the server's timezone.

Channel Outputs

When an engine completes a run, the output can be delivered to one or more channels:

  • Slack: Post the result to a Slack channel via the Slack connector
  • Email: Send the result as an email to specified recipients
  • Webhook: POST the result to a custom URL

Outputs are configured during engine creation and can be updated at any time.

Run History

Every engine execution is logged in the run history. Each entry includes:

  • Timestamp: When the run started
  • Duration: How long the run took
  • Status: Success, failure, or timeout
  • Token usage: Input and output tokens consumed
  • Output: The generated content
  • Error details: If the run failed, what went wrong

Run history is accessible from the engine detail view and is also recorded in the audit log.

Manual Triggers

In addition to the cron schedule, any engine can be triggered manually by clicking the Run Now button in the engine detail view. This is useful for:

  • Testing a new engine before its first scheduled run
  • Running an on-demand report outside the regular schedule
  • Debugging a failing engine

Token Tracking

Each engine run tracks token usage separately, so you can see how much each automation costs over time. Token data includes:

  • Input tokens (the prompt and context sent to the model)
  • Output tokens (the response generated by the model)
  • Estimated cost based on the configured model rates

Example: Weekly Sales Summary

Here's an example engine configuration:

  • Name: Weekly Sales Summary
  • Schedule: 0 9 * * 1 (Monday 9 AM)
  • Timezone: America/New_York
  • System prompt:
You are a sales analyst. Using the connected PostgreSQL database:

1. Query the orders table for the past 7 days
2. Calculate total revenue, order count, and average order value
3. Compare with the previous week
4. Identify top 5 products by revenue
5. Format as a Slack-friendly summary with key metrics bolded
  • Tools: PostgreSQL
  • Output: Slack (#sales-team channel)

Next Steps