Introduction on Grafana Labs

Plan a reliable daily summary

Sat, 17 Oct 2026 02:03:35 +0100

A summary worth repeating

You review the checkout service each morning and need a summary of request volume, error rate, and notable errors. Repeating that question manually takes time, but repeating an unclear prompt can produce an unhelpful summary every day.

This journey is for Grafana Cloud users who already ask Assistant questions about their data and want to plan recurring, read-only work.

What you’ll learn

Identify a useful recurring task and the context its prompt needs.
Explain who can access the saved automation and whose permissions its runs use.
Evaluate a run’s evidence and choose a schedule that matches when you need the result.

Understand Automations

Sat, 17 Oct 2026 02:03:35 +0100

From one question to recurring work

An automation is a saved Grafana Assistant prompt that runs on demand or on a schedule. Each run creates a dedicated Assistant conversation containing its results.

This journey covers saved-prompt automations, where each run creates a separate conversation. Investigation automations repeat checks within an existing investigation chat.

For the checkout summary, the saved prompt defines the service, data sources, and reporting window. A schedule determines when Assistant repeats that work.

A useful starting task has a result you want every time it runs, such as a weekday service summary. Keeping the task read-only lets you evaluate the findings without asking Assistant to change resources.

Before you begin

Sat, 17 Oct 2026 02:03:35 +0100

Access and data for your summary

Automations are available only in Grafana Cloud. They aren’t available in self-managed Grafana, even when connected to a Grafana Cloud Assistant backend. To apply these concepts, you need:

Assistant available in your stack, with any required terms accepted.
Permission to access Assistant and manage personal automations.
A service with data you can query, and access to its data sources.
Available Assistant usage for manual tests and recurring runs.

Automation permissions don’t grant access to the underlying data. If either is missing, an administrator can help with the required permissions.

The checkout service and data source names in the examples are illustrative. Your own summary needs the names and data available in your stack.