Scheduled monitoring jobs
Turn frequent questions into scheduled jobs that run automatically on a set schedule, push results to IM groups, and log failures automatically.
Create a job
Create a scheduled monitoring job inside a Space, providing:
- Name: e.g. "Daily operating report";
- Question: the same natural-language question as in chat, e.g. "yesterday's GMV by channel with change vs. prior";
- Schedule: pick a frequency and time, or use a cron expression for precise control;
- Push channels: DingTalk / Lark / WeCom, multi-select;
- Tier: the model tier for this job (defaults to the Space's default if blank).
A job belongs to a Space and reuses that Space's datasets and business definitions.
Schedule
Use a simple frequency (e.g. daily at a given time) or a cron expression for precision. Common examples:
| Need | cron |
|---|---|
| Daily at 09:00 | 0 9 * * * |
| Mondays at 09:00 | 0 9 * * 1 |
| Hourly | 0 * * * * |
| Every 15 minutes | */15 * * * * |
Run history
After each run, the job records the last / next run time, success or failure, and the error on failure — handy for troubleshooting. You can also open the corresponding conversation to see the full result.
Push to IM groups
Results are pushed to the channels you selected. First configure the platform's bot under IM integrations. If one channel fails, the reason is logged separately without affecting the others.
Run now
You can manually "Run now" once to verify, and immediately see the result and each channel's push status.