Data Visualizer

Some numbers are easy to read. A table of twelve months of revenue is not. Your eye wants a line that climbs or dips, or a bar that is clearly longer than the others. The Data Visualizer gives you that picture. It takes the findings and draws one chart for each key point, then builds a dashboard you can open and explore.

Think of it as the illustrator on a magazine team. The writer has the story, and the illustrator makes sure one picture carries the point at a glance. Every chart gets a title that states the takeaway, and both axes are labelled, so the reader never has to guess what they are looking at.

It also checks its own work, like a proofreader before the page goes to print. It looks at every chart and dashboard before passing them on, and it draws only data computed from your files. It never adds an invented line to make a point look stronger.

Think of a weather map on the evening news. Nobody reads a table of temperatures for a whole country. They look at the colours and the lines and see the story at once. A good chart does the same job: the reader sees the pattern in a second, then looks closer if they want the detail.

• Make a chart for each key finding: trends over time, top-ten lists, shares of a whole, spreads, relationships and comparisons.
• Choose the chart type that fits the finding, from line, bar and donut charts to heatmaps, scatter plots, histograms and box plots.
• Title every chart with its takeaway, such as "Revenue grew this quarter", and label both axes.
• Build an interactive dashboard with KPI cards, a grid of charts and a data table, and open it in your browser.
• Keep the charts in step with the report, so every chart the report mentions actually exists.
• Remember the chart types, colours and dashboard layout you prefer, and use them next time, so every new chart in your project looks like part of the same set.
• Keep dashboards safe to share. A dashboard loads only its charting library, with no trackers and no outside content.
• Work with your connected apps once you connect them: Google Sheets, Google Slides and Google Analytics.

It works in three panels: the files panel for your charts and data, the browser panel for previewing a dashboard, and the artifact panel, where a dashboard opens so you can share it. It reads your data in place and writes its charts only to the project's own output folder. It never changes the data it was given.

The Data Visualizer has three skills. You can ask for any of them in plain words.

visualize makes publication-ready charts from your data, in the team's chart style. It covers line, bar, donut, heatmap, scatter, histogram and box plot charts. Use it for "Show me monthly revenue as a chart."

dashboard builds a standalone, interactive HTML page with KPI cards, charts and a data table. You open it in your browser, and you can share the file. Use it for "Build a dashboard I can open in the browser."

pipeline-checks is a set of optional checks that run before and after an analysis: a data check before the run, a log of past runs, and a webhook notice when it finishes. Each one runs only when you approve it. Use it when you want a safety check before a chart run that feeds a live page.

Here is a common job, from a question to a dashboard you can share.

1. You ask: "Build a dashboard I can open in the browser for weekly sales by region."
2. The Data Visualizer reads the clean data and the findings from the Data Scientist. If a finding it needs is missing, it says so and does not draw a stand-in.
3. It picks a chart for each finding. A line shows the weekly trend, and a bar chart ranks the regions.
4. It writes each chart title as a takeaway, labels both axes, and groups small categories into "Other" when there are too many to read.
5. It asks which look you want from the themes it keeps, or whether the default is fine.
6. It assembles the dashboard: KPI cards along the top, the grid of charts in the middle, and the data table underneath.
7. It opens the result and looks at each chart. If a label overlaps or a scale misleads, it fixes it before handing the page on.
8. You get a single dashboard file in the artifact panel. You can open it in your browser, and it loads nothing except its charting library.

At the end, you have one dashboard file, a set of chart files each with a takeaway title, and a note of any finding that was missing or left out. Your data is unchanged, and every chart is one the report can refer to. Anything hard to undo waits for your yes.

Call it when a number has to travel. A finding that lives in a spreadsheet reaches the people who open spreadsheets. The same finding as a titled chart reaches everyone in the room, including the person who only has thirty seconds. Ask for it before a review meeting, a weekly update or an investor note, and whenever you catch yourself explaining the same table twice.

Ask it also when you are not sure what the data is saying yet. A quick set of charts is often the fastest way to see the shape of a problem: one region falling while the rest climb, one product carrying the quarter, a seasonal dip that looks like a crisis until the line from last year sits beside it. It draws only what the data supports, so the picture you react to is the real one. If a chart would mislead, say a scale that hides a drop or a comparison with too few points, it tells you and offers a better view instead.