Skip to main content
A dataset is a spreadsheet entry. It opens as a full-width grid where you edit cells directly, write formulas, and set up filters and sorting — and artifacts can read its data live (see Live data in artifacts).

Creating a dataset

There are three ways to create one:
  • Import a file. Import an Excel or CSV file — each sheet in an Excel file becomes its own dataset, and the original file is kept so you can download it later. See Importing files.
  • Start blank. Click New in the top bar, choose More types…, and pick Dataset.
  • Ask the agent. Describe the data you want and the agent creates and fills the dataset for you. It can also look up answers, add rows, and update cells in an existing one.

The grid

Click any cell to edit it directly, or use the formula bar above the grid. A new dataset starts at 26 columns (A–Z) and 1,000 rows and grows as you need:
  • At the bottom of the grid, Add 1000 more rows at the bottom extends it — change the number first if you want more or fewer.
  • Insert column right and Insert row below add a column or row next to the selection.
Selecting a whole row or column opens its toolbar, with Bold, Cut, Copy, Paste, clearing the values, inserting, and deleting. Drag rows and columns to reorder them. You can also drag the fill handle to extend a value or formula across neighboring cells, and apply a Fill color to cells. When a cell’s text is too long to fit, rest the pointer on it to read the whole value.

Labels and totals

  • Layout, in the dataset’s Fields panel, says where the labels are: in the first row, the first column, or neither. Studiograph guesses when you first open the dataset; change it if the guess is wrong.
  • Pin to bottom, on a row’s toolbar, keeps a row — a totals row, say — at the bottom of the grid, outside sorting and filters.

Formulas

A formula starts with =, like a spreadsheet:
  • =A2*B2 — arithmetic between cells
  • =SUM(C2:C10) — a function over a range
Supported functions are SUM, AVERAGE, MIN, MAX, COUNT, and PRODUCT, plus arithmetic and date math. Suggestions appear as you type after =. The formula itself is stored in the cell; the grid shows the computed result, and the formula bar shows the formula. Functions over a range include rows hidden by a filter, like SUM in Excel. A formula that can’t be computed shows an error value: #REF! (a cell reference that no longer exists), #DIV/0! (division by zero), #CYCLE! (a formula that depends on itself), #NAME? (an unknown function), #VALUE! (the wrong kind of value), or #ERR.

Column formats

Set a column’s format to control how values display: Text, Number, Currency, Percentage, or Date, with choices for the currency, the Date format, and the number of Decimal places. Some formats also tidy the values themselves when you apply them:
  • Date rewrites dates it can read into the standard YYYY-MM-DD form.
  • Number and Currency turn text like $1,234.00 into plain numbers.
  • Percentage changes only how values look.
You can undo a format change like any other edit.

Filters and sorting

Each column can have a filter suited to its type:
  • Numbers — Equals, Not equals, Greater than, Less than, At least, At most, or Between.
  • Dates — On, Before, After, or Between.
  • Text — Contains, Equals, Starts with, Ends with, or Is empty.
  • Values — tick the values to keep, with Select all and Deselect all.
Every filter has Clear filter. The dataset can also have one sort: click the sort icon on a column to cycle through ascending, descending, and off. The sort follows its column if you reorder or rename columns.
Sort, filters, column formats, column widths, and the layout are shared view settings: everyone who opens the dataset sees the same setup. Changing a filter changes it for the whole team, not only your own view.

Working together

Teammates can edit the same dataset at the same time and see each other’s changes live. To discuss a value, select the cell and choose Comment — commented cells are tinted so they’re easy to spot. The History panel shows earlier versions of the dataset, and Restore brings one back.

Downloading and sharing

Open the … menu and choose Download:
  • CSV (.csv) — the whole dataset, with formulas turned into their computed values. It includes every row in its stored order, regardless of the current filters and sort.
  • Markdown (.md) — the dataset’s source file, with formulas kept as written.
  • Original file — for an imported dataset, the spreadsheet it came from.
A public link shows the dataset as a read-only sheet, with the team’s filters, sort, and formats applied.

Limits