Skip to main content
AI models are good with words and unreliable with arithmetic. Ask one to add up 400 order totals and it will give you a confident number. It might even be right. Arora doesn’t leave that to chance. When the data is on your workspace, the server does the math. Same input, same answer, every time.

​
Count

Count lines, words, characters, or how many times something appears.

​
Totals and averages

Sum, average, minimum, maximum, or median of the numbers that follow a label you name. Dollar signs and commas are fine. Amounts in parentheses count as negative, the way accountants write them.
The answer comes back with a short audit: how many labels the server found and how many numbers it pulled. If those don’t match, some rows were missing a number, and your AI can tell you which.

​
Compare two versions

A line-by-line diff of two documents, or of a document and a new draft.

​
Charts, diagrams, and equations

Your AI writes the chart or diagram. The server turns it into a hosted image you can drop into a doc, an email, or a Slack message.
  • Charts from your data: bar, line, pie, and more.
  • Diagrams and graphics like org charts, flows, and simple layouts.
  • Equations rendered cleanly, for pricing formulas or commission math.
Charts use the data your AI hands over directly. They can’t pull from an outside link, so ask your AI to compute the numbers first, then chart them.
These are actions on bridge_idavoll: count, aggregate (sum, avg, min, max, median, count), diff, and render. render accepts SVG, Vega-Lite JSON with inline values, or a LaTeX equation, and returns a hosted PNG URL. Fonts available for SVG text: Inter, Outfit, Source Serif 4, and JetBrains Mono. Stage a large spec on the workspace and render it by name; inline specs over about 16 KB are refused.
Last modified on October 13, 2026