Charts an AI built for you are real, running JavaScript. Publishing them as a page keeps them interactive instead of flattening them into a screenshot.
Charts and tooltips keep working — publish it interactive, not as a screenshot.
Ask an AI assistant to "turn this CSV into a dashboard" and you get back a single HTML file with a charting library, your data inlined, and some tooltips. It works in the preview panel. Then you need to show it in Monday’s meeting, and the usual fallback is a screenshot — which throws away every hover, filter, and drill-down that made it a dashboard rather than a picture of one. Publishing the file as a page keeps the interactivity, and gives you something to paste into a calendar invite so people can look before the meeting instead of during it.
Copy the output straight from your AI tool — the raw HTML or markdown, not a screenshot. There is no file to save and no app to install first.
Paste it into the box above. The preview renders it the way the recipient will see it, so you can confirm the formatting before you share.
Create the link and send it. Whoever you send it to opens a clean, read-only page in their browser — on a laptop or a phone — with nothing else from your chat attached.
An operations lead exports last quarter’s ticket data and asks Claude for a dashboard: volume over time, a breakdown by category, and median resolution time, with the data embedded so the file stands alone. The result is about 400 lines of HTML with a charting library pulled from a CDN. Pasted and published, the charts render and the tooltips respond, because scripts run in the published sandbox and an external library loaded over HTTPS still loads. What does not carry over is anything reading from a local file path — a dashboard written to fetch("./data.csv") shows empty charts, because there is no sibling file next to a published single document. The fix is to ask for the data inlined in the HTML, which is worth specifying in the original prompt. Since the numbers are internal, the lead adds a password before sending the link to the team channel.
Yes, provided the library is fetched over HTTPS from a public URL, which is how AI-generated dashboards normally reference one. The published page runs scripts inside a sandboxed frame, so the library loads and draws as it did in your preview. What fails is a reference to a file that only exists on your own machine.
Almost always because the dashboard reads its data from a separate local file rather than carrying it inside the HTML. A published document is one standalone file with no folder around it, so a relative fetch has nothing to resolve against. Ask your AI tool to regenerate it with the dataset embedded directly in the page and the charts fill in.
No, and it should not pretend to be. There is no scheduled refresh, no live database connection, and no row-level access control — the page is a snapshot of the data that was in the file when you published it. It is a good fit for a one-off analysis, a board slide, or a shared look at last quarter, and a bad fit for a metric anyone needs to watch continuously.
Every published page carries noindex directives, so a dashboard never enters a search index no matter which plan you are on, and the free tier’s seven-day expiry means a sensitive snapshot stops existing without anyone having to remember it. To go further and require a password before the dashboard opens at all, that sits on a paid plan.
免費帳號的連結保留 7 天。升級付費方案後，你建立的每個頁面都會永久上線。
查看方案與價格