Anthropic Claude Dashboards: What Your Data Must Supply
By Alexandre Saint-Jean
Audio version
Audio version produced by text-to-speech from the article. Our AI charter
On 8 October 2026, Anthropic published a page (Claude, 8 October 2026) presenting Claude Dashboards, a beta feature on paid Claude plans that turns company data into a dashboard that stays up to date. For a finance or HR manager still steering indicators in Excel, two questions matter: what the tool does, and what you must supply for the number on screen to be right.
What does Claude Dashboards actually do?
According to Anthropic's page, Dashboards turns business data into a dashboard that stays up to date, and each chart shows when its data was last refreshed. Anthropic presents it as suited to quick, exploratory questions. Deeper analysis is handed to BI (business intelligence) tools.
Our take: it is a first look at the data, not a full reporting platform.
Which data sources can it connect to?
Anthropic names Amazon Redshift, BigQuery, ClickHouse, Databricks and Snowflake, plus CRM tools such as Salesforce. The page adds that Dashboards also works with other connectors. That list is the first thing to compare with where your figures actually live today.
Our take: the sources Anthropic names do not include a spreadsheet. A team whose indicators sit in Excel files should ask, before anything else, whether its data is reachable through one of the other connectors, or whether it first needs to move somewhere the tool can read.
One condition applies before any test: Dashboards is off by default, and administrators enable it under Organization settings > Artifacts.
How can you check a number before you quote it?
Anthropic says you can click any number to see the query behind it, and ask Claude to explain it. In practice, a manager can verify where an indicator comes from before citing it in a report or a budget review. The last-refresh time on each chart answers a second question: how old is the figure you are about to quote?
Our take: the query shows how a number was computed, not whether it is the number your business means. Reading it well takes someone who knows what "headcount" or "revenue" should include in your company. The tool makes the check possible; it does not do the judging for you.
What must you define before the dashboard is right?
A dashboard generated by AI raises the same question as an AI agent working on company data: which data does it rely on, and who decided? We cover that in our article on how AI agents use company data. Three decisions belong to you, not to the tool.
Our take: name one authoritative source for each indicator. If sales appear in both the CRM and the accounting file, and the two disagree, a dashboard built on the wrong one will look equally tidy. Decide in advance which source counts, and write it down before anyone connects anything.
Our take: fix the filters. Period, entity, currency, contract type, active or departed staff: each choice changes the figure. Write them next to the indicator, so the person reading the chart knows what is in and what is out, and so anyone asking Claude for a new view starts from the same scope.
Our take: give each indicator a one-line definition. "Headcount" may mean contracts, people or full-time equivalents; "revenue" may mean invoiced or collected. Claude can show the query, but only you can say which meaning your company has settled on.
When is a Claude dashboard the wrong tool?
Anthropic describes Dashboards as suited to quick, exploratory questions. For deeper analysis it hands off to BI tools: Amplitude, Grafana, Hex, Mixpanel, Omni, Perplexity, PostHog and Sigma. Databricks Genie, Looker, monday.com and Tableau appear in the "coming soon" list on the page.
Our take: if your reporting already runs on Power BI, compare a Claude dashboard with it on a single indicator before changing anything. Our Power BI Copilot guide for SMEs covers the other side of that comparison.
Should you try Claude Dashboards now?
Our take: yes, on one precise question, if your data sits in a source Anthropic names or behind another connector and an administrator can switch the feature on. Compare the result with your reference figure before relying on it. If your indicators live in Excel and nobody has written their definitions, that is the useful work, with or without a tool.
Frequently asked questions
- What is Claude Dashboards?
- According to Anthropic's page dated 8 October 2026, Claude Dashboards is a beta feature on paid Claude plans that turns company data into a dashboard that stays up to date. Each chart shows when its data was last refreshed, and any number can be clicked to see the query behind it.
- Which data sources does Claude Dashboards connect to?
- Anthropic names Amazon Redshift, BigQuery, ClickHouse, Databricks and Snowflake, as well as CRM tools such as Salesforce. The same page says Dashboards also works with other connectors, so check with your Claude administrator before planning around it.
- Can you trust the numbers in a Claude dashboard?
- You can check them: every number links to its query, and Claude can explain it. Our take: whether a number is the one your business means depends on the source you connected, the filters applied and the definition of the indicator, and those remain your decisions.
- Is Claude Dashboards switched on by default?
- No. Anthropic states that Dashboards and Motion are off by default, and that administrators enable them under Organization settings > Artifacts. An employee on a paid plan may therefore not see the feature until an administrator acts.
Sources
Go further
AI agents and company data context