Schedule and deliver an analysis

Schedule a checked Explore, choose email or Slack delivery, and verify the result.

Use an automation to refresh an Explore regularly and send its notebook results to your team. Start with work you have already checked: a schedule will repeat the analysis, including any mistake in it.

Before you start

Check the Explore's code and results. If the answer is uncertain, verify it first.
Have workspace file-write access to save the automation. Make sure the person responsible for scheduled execution can run the analysis: queries require datasource permission and exactly one authorized credential. Automation ownership grants neither. See credential selection.
Decide who may receive the rendered results. Email addresses and Slack channels can expose tables and chart images outside the Explore's access controls.

1. Set the schedule

Open Automations in the workspace sidebar and select New Automation. Give it a recognizable title, such as Daily checked analysis.

Select the Schedule node. For a daily run at 09:00 UTC, set Run every to Day and At to 09:00. Check that the displayed timezone is UTC and Cron is 0 9 * * *.

The simple controls also offer Hour, Week and Month. Select Use cron expression for a custom five-field schedule. The builder shows the stored timezone but has no timezone picker. Its Next run preview does not apply that timezone, so check the cron and timezone together when choosing the intended time. See Cron and timezones.

2. Choose what runs

Select Add workflow, then choose:

Workflow	Use it when	Configure
Explore sync	You want to refresh the checked Explore's existing notebook.	Select it in Explore and confirm the right title. Its datasource and query variables come from that notebook.
Run a prompt	You want a new Explore for each run.	Fill Prompt, Model and Datasources, with at least one datasource. Ask for findings in notebook cells; an agent's chat reply alone is not delivered.

For a recurring report based on work you have checked, use Explore sync. It refreshes that Explore rather than starting a new analysis each time.

3. Choose delivery

Leave out the action if you only want the analysis to run. To send results, select Add action and choose one destination type:

Action	What to do
Send email	Add approved addresses in Recipients. Press Enter for each one and check the resulting chips.
Slack message	Choose a channel in Send to. If you see Configure Slack first, connect the workspace's Slack installation before continuing.

One automation can have one delivery action. Email supports several addresses; Slack targets one channel. Email recipients do not have to be workspace users. Review the actual recipients and the data they will see before testing or publishing. A delivery address does not grant a query credential or permission to open the Explore.

Delivery uses rendered notebook output. If layout cells are selected, only selected cells present in the run result are included. Check that the intended tables and charts are in the notebook and layout; without renderable output, no email or Slack message is sent. The View in Querio link also does not grant access. Share the Explore with people who need to open it.

4. Publish and test

Review the workflow, schedule, timezone and destination, then select Publish. The automation becomes Live, which enables recurring execution and its configured delivery. Draft pauses automatic scheduling.

Open the saved automation to use Test run; the new-automation screen does not show it. Test run saves the current definition and starts real execution and delivery. It is available for a draft too, without publishing it. Check recipients first: a test can refresh the notebook or create an Explore, then send email or post to Slack.

A manual test runs as its caller. Scheduled execution uses the recorded creator attribution, so a test by someone else does not establish that the scheduled run has the same access. See Workflows and run ownership.

Automatic scheduling reads canonical automations/*.yaml files under the configured context root. Publish working-copy edits before relying on a changed schedule. A successful manual test does not establish that the schedule will run. See Definition and published state.

5. Verify the run and the message

Open Runs and find the run marked Test. A run ID means the request was accepted; wait for the execution result and inspect Details.

When execution finishes, use View Explore to check the actual analysis against the result you expected.
If you configured delivery, check the inbox or Slack channel. Inspect the values and images, not just whether a message appeared.
If recipients need to open the Explore, have one use View in Querio with their own account.

Completed means the analysis executed successfully. It does not confirm delivery, and Runs has no separate delivery receipt. A message may also arrive with a missing chart. Check the destination before retrying: another test repeats the analysis and may send a duplicate. For delivery limits and failure states, see Automations.

After the next scheduled time, check Runs for a scheduled execution and verify its result and destination again. The earlier test and Next run preview do not prove recurring scheduling worked.

Change or pause it

Open the automation and select Unpublish. It becomes Draft, and you can edit its builder controls. Review the changes, then select Publish to resume scheduling.

Unpublishing stops future scheduling once the scheduler picks up the change. It does not cancel work already started or recall a message. A run keeps the delivery settings captured when it was submitted, even if you change the destination afterward.

If a run fails or stays pending, inspect its Details and any Explore it produced before retrying. Use the automation reference for failure and recovery steps.