Create a dashboard

Use this page to translate a monitoring idea into a working Grafana dashboard with help from the Assistant. The workflow keeps you in control while the Assistant handles navigation, panel creation, and query scaffolding.

What you’ll achieve

Capture requirements for a new dashboard and communicate them clearly to the Assistant.
Review and refine generated panels, queries, and visualizations before publishing.
Share the finished dashboard with teammates or use it as a template for other environments.

Before you begin

Clear monitoring goal: Know the service, feature, or SLO you want to visualize.
Data source access: Verify the Assistant can query the metrics, logs, or traces you intend to use.
Folder path: Decide where the new dashboard should live and whether additional reviewers need access.

Draft the dashboard structure

Explain the monitoring goal to the Assistant and outline the panels you expect so it can scaffold the dashboard with the right context.

Open Grafana Assistant and describe the outcome you need.
Break the request into small chunks, such as panel count, target metrics, groupings, and alerting requirements, so the Assistant can address each part in sequence.

Create a dashboard called "Checkout health" with five panels: error rate, p95 latency, request volume, 5xx logs by status, and active carts from @orders-database.

Mention the target data source for each signal. The Assistant uses your hints to call the right tooling and generate panels. Include examples or existing dashboards when you want it to mirror a proven layout.

Review the generated layout

Walk through each newly created panel to confirm queries, visualizations, and layout decisions match your expectations.

The Assistant navigates to /dashboard/new, sets the title, and creates the requested panels.
Inspect each panel for correct queries, visualization types, and display options.
Ask follow-up questions if anything looks off and correct the Assistant politely when details are wrong:

Switch the error rate panel to a bar gauge with a 5-minute interval.

Add a query that breaks request latency down by region.

Capture the dashboard intent, persist the draft, and notify collaborators so they can start using it.

Update the dashboard description with your monitoring goal so teammates understand the intent.
Click Save dashboard and choose the destination folder.
Share the dashboard link in your team channel and mention any next steps, such as alert creation or runbook updates.

You now have a saved dashboard seeded with Assistant-generated panels. Review the dashboard with stakeholders, capture feedback, and iterate using the related workflows below.

Troubleshooting

Data source mismatch: re-run the prompt with explicit @datasource mentions for each panel.
Empty panel: ask the Assistant to validate the generated query, for example, Validate the query in the latency panel and fix any syntax issues.
Layout needs tweaking: follow the Add variables and adjust layouts guide to refine positioning and reuse filters.