Cloudflare Radar visualizes worldwide Internet traffic patterns, attack activity, and technology trends, drawing on data from Cloudflare's global network and from aggregated, anonymized queries to the 1.1.1.1 public DNS Resolver. It also exposes a publicly accessible API, which lets users go beyond the default charts — filtering by protocol, comparing metrics across locations or autonomous systems, or examining trends over two separate time periods. The barrier is that not every user can hand-write a raw API query or work through a JSON response. Two new additions address that gap: the Cloudflare Radar Data Explorer, a web interface for assembling complex queries (comparisons and filters included) and charting the output, and an AI Assistant built on Cloudflare Workers AI that turns plain-language questions into the corresponding Radar API calls, with the results rendered in the Data Explorer.
Submitting a question in plain language
Queries built from dropdowns and filters can be overwhelming when a user knows what they want but not how to express it through the available options. In that situation they can put the question to the AI Assistant directly — for instance, "Has there been an uptick in malicious email over the last week?" The assistant makes a sequence of Workers AI and Radar API calls to fetch the data and produce a visualization within seconds.

On desktop browsers the assistant sits in a pane on the right side of the page; on mobile it appears when the user taps the "AI Assistant" button. Questions are typed into the "Ask me something" field at the bottom of the pane and submitted. Sample queries shown by default illustrate what can be asked, and clicking one submits it.

After evaluation, the visualization lands in the Results section of the Data Explorer, and the matching Data, Filter, and Compare selections appear in the Query section above it, so results can be tuned further by hand. The Keep current filters toggle lets a user stack questions: with it active, asking "Traffic in the United States" and then "Compare it with traffic in Mexico" adds Mexico to the existing graph.
Assembling queries manually
Users who prefer direct control can explore datasets covering traffic metrics, attacks, Internet quality, email security, and more. Selecting a dataset automatically populates the Breakdown By: dropdown with the relevant options where they exist, along with dynamic Filter choices, and the visualization in the Result section updates as selections change.
Comparisons are built into the same interface. The Previous period switch compares against the immediately preceding window (seven days against the seven before, for example). Alternatively, the Date Range field opens a calendar where picking a start date triggers an intelligently chosen range based on the Filter selection. To compare across locations or autonomous systems, the "Location or ASN" field offers a search box that accepts a country or region name, an AS name, or an AS number, with results updating as the user types. Locations can be compared with other locations or with ASes, and ASes with other ASes or locations — enough, for instance, to set an ISP's trends against those of its country.

How results are drawn
Data Explorer chooses among a time series graph, bar chart, treemap, and global choropleth map based on what the API returns. Country alpha2 keys in the response produce a choropleth map; timestamps in the response produce a line graph ("xychart"); more than 40 items in the response select a treemap.
An expanded version of a chart normally found on Radar's Adoption & Usage page shows what the extra controls buy. The standard graph plots HTTP version shares for the United States over the last seven days alongside summary shares. In this extended Data Explorer version, the same HTTP version share data is compared between the United States and AS701 (Verizon), across both the past seven days and the previous seven-day period, with the summary values compared in an accompanying bar chart.

Dense time series can carry many plotted lines. Double-clicking an item in the legend isolates that single line; single-clicking a legend item adds or removes the corresponding line.



