Vireo

Getting started

Vireo is a spreadsheet where the AI assistant does the building and you do the checking. Bring in a file and describe what you need, say a chart or a valuation, and the assistant assembles it as a visible pipeline of steps you can inspect one at a time. This page walks you from a raw CSV to your first assistant-built result.

Import your data

Vireo reads CSV and Excel files. Use File > Import, or drop the file into the window. For Excel files, Vireo finds the actual data region on the sheet, skipping title rows and notes, and shows you a preview before anything is imported. Check there that the headers and types look right.

Real-world files are messy, and Vireo expects that: fiscal quarters in column headers, negatives in parentheses, currency prefixes, thousands separators. It reads them as the numbers and dates they are.

Tables are the workspace

Each imported file becomes a table on its own tab, like a sheet in a workbook. Every column has a type (number, text, or date), and Vireo detects the type on import. Types matter: a Close_Date column typed as dates can be grouped by month; the same column typed as text is plain strings. If a column comes in wrong, right-click its header and change the type.

You can edit cells directly, and every edit, yours or the assistant's, lands on the undo stack.

Formula columns handle the row-level math

For a calculation that runs down every row, add a formula column. Column names go in square brackets:

=[Revenue] - [Costs]

gives you a margin column that recalculates whenever the inputs change. The built-in functions cover the Excel staples: IF, SUMIF, VLOOKUP, PMT, NPV, IRR and the rest. The full list is in the formula reference.

Formula columns are for arithmetic within a row. For anything structural (filtering, grouping, joining tables, forecasting), ask the assistant.

Ask the assistant to build for you

The chat panel sits alongside your sheet. Describe what you want in plain language, naming the table and columns where you can:

The assistant builds the steps in front of you and reports back in chat with a summary of what it created. If the result isn't quite right, say so: "exclude the EMEA region" or "make that quarterly instead of monthly". The assistant adjusts what it built rather than starting over.

Every step is a node you can inspect

The assistant works on a canvas next to the sheet, assembling transformations from building blocks called nodes: Filter, GroupBy, Join, Pivot, ExtendForecast, LineChart, and about 70 more. Nodes connect into a pipeline: your table flows in one end, the result comes out the other.

Nothing the assistant builds is a black box. Click any node to see what feeds into it, what its settings are, and what it currently produces. If a step breaks, Vireo points at that node and says why. The node reference describes every building block.

You can also build or edit pipelines yourself: drag a table onto the canvas and wire it through nodes, or adjust a setting the assistant chose. The sheet and the pipeline are the same data; there is no separate export step between them.

Results come back as tabs

When a pipeline produces a table (a filtered subset, say, or a forecast extended five years out), the result appears as a new tab in the workbook, next to your source data. Edit a cell in the source and the pipeline re-evaluates; the result tab updates to match. Charts and KPI cards the assistant builds show up on the canvas and can be assembled into dashboard forms.

Save your work

Your workbook saves as a .tgraph file: tables, pipelines, formulas, and the assistant's full history in one file, on your machine. There is no cloud copy and no account; the file goes where you put it. Vireo can also auto-save on an interval if you turn that on.

From here, the natural next step is the assistant guide: what the AI can take on and how to check its work.