What are Claude Dashboards? How to analyze and visualize data with Anthropic's Claude.ai

Short answer

A Claude Dashboard is an artifact, a page Claude builds next to your chat, connected to a real data source. Each chart runs its own SQL query, shows it to you and says when its data was last refreshed. You refine it by talking to Claude, not by dragging fields around a BI tool.

Ask a business question. Get charts you can check, query by query.

How Claude Dashboards work

• Connect a source: data platforms such as Amazon Redshift, BigQuery, ClickHouse, Databricks and Snowflake, or another connector such as Salesforce
• Ask in plain language: request a dashboard in any chat, pick Output > Dashboards in the message box, or start from a Dashboards template in the Artifacts tab
• Claude writes and runs a SQL query for each chart, then arranges the results into KPIs, tables and charts
• Every chart shows its query and its last refresh time, so you can see exactly what is being counted
• Refine it in the chat: add a filter, switch to a weekly view, or open a query and tell Claude what to fix

Who can use Claude Dashboards?

The beta is available on the paid Pro, Max, Team and Enterprise plans. On Enterprise it is off by default until an owner enables it under Organization settings > Artifacts. Dashboards start private, can be shared with the Share button and count toward your plan's usage limits.

What Claude Dashboards are good at

Anthropic positions Dashboards for quick, exploratory questions: how this week's signups compare with last month's, which region is behind on revenue, where the sales pipeline stalls. A question that used to wait in an analyst's queue becomes a chart in minutes. For deeper analysis, you can send a dashboard on to tools such as Hex, Mixpanel, Grafana or Sigma.

Limits to keep in mind

• It is a beta: features can change, and Anthropic notes that some betas never reach general availability
• The numbers are only as good as the query: Claude writes SQL from your wording, so define terms like “active customer” or “revenue” precisely
• Check the SQL before a chart drives a decision
• It does not replace a governed BI layer, or a product dashboard your customers use

How to get better results

Describe the decision, not the chart type: “Which product lines lost margin this quarter?” beats “make a bar chart”. Name the table or app, state the time range and how each metric is defined, then review the first version chart by chart.

Claude Dashboards vs a custom dashboard

Claude Dashboards suit internal questions from people who already have access to the data. A custom dashboard still makes sense when customers or partners need to see it, when the logic must be identical every time, or when the data lives in systems Claude cannot reach. At Athryx Tech we build that second kind, inside custom software and mobile apps.

Conclusion

Claude Dashboards turn “can someone pull the numbers?” into a conversation. Use them for fast exploration, check the queries before you decide, and keep your key metric definitions written down.