The work an analyst does before the first chart.
Reads the file, not the filename
Types every column from its values, not its header. Finds the real header under title and blank rows. Hundreds of column aliases across 36 business domains.
Knows what your industry measures
157 metrics with definitions, not just column sums. Margin for sales, tenure for HR, RevPAR for property. Skips any metric your columns cannot honestly support.
Picks the chart the data calls for
Shape follows the column: histogram, bar, donut, scatter. A time series whenever a date column exists. Click any bar to see the rows behind the number.
Explains itself in one sentence
“Revenue rose 9.4% while order value fell”, not “chart shows revenue”. Labelled written or computed, so you always know which. Generated when you ask, so nothing is billed unread.
Does not fall over on real files
50,000 rows sorted and searched server-side. Corrections write a new version; undo is real. 100 MB files, and Excel read 5.4× faster than openpyxl.
Built to survive a security review
Your rows never leave the process. Only column names do. Every dashboard works with no AI configured at all. Export as PDF, WebP or one self-contained HTML file.