> ## Documentation index
> Fetch the complete documentation index at: [llms.txt](../llms.txt)
> Use this file to discover all available pages before exploring further.

# Datasets: rows, columns, and cells

What a Sleuth dataset is, which cells use credits, and how to change one.

[Open Analyze](https://app.sleuth.studio)

Every analysis in Sleuth sits on a dataset: a table where each row is a record, like a drug program, a company, or a trial, and each column answers part of your question.

## Two kinds of cells

| Kind | Example | Credits |
| --- | --- | --- |
| **Fact cells**: values Sleuth's knowledge base already has | An asset's sponsor, target, modality, and highest phase | Free |
| **Analytical cells**: judgments Sleuth makes for your question | How well an asset fits your strategy, or whether a program has white space in 2L NSCLC | 1 credit each |

So scoring 150 assets on 3 criteria uses 450 credits, while their phase and sponsor cost nothing. [How credits work](how-credits-work.md)

## What's in every cell

Click any cell to see its value, the source it came from, the passage behind it, and a confidence score. Choose **See more** for the score breakdown and Sleuth's explanation. [Confidence scores](confidence.md)

## Change a dataset

- **Edit a value** by double-clicking the cell. Your correction gets a confidence of 100.
- **Re-run a cell** with new guidance, like "use the label, not the press release."
- **Add rows or columns** by asking in the chat, or with **Add**. New AI-filled columns are custom columns. [Add your own columns](custom-columns.md)
- **Delete** rows or columns you don't need.

You can't edit a past version of a dataset, or a dataset someone else owns.

## Versions

When monitoring or an edit changes a dataset, Sleuth saves a new version with a log of what changed. You can open any earlier version. [Monitoring](monitoring.md)

## Download it

Download any dataset as a CSV that opens in Excel. [Outputs and exports](outputs.md)

## Related

- [Add your own columns](custom-columns.md): Ask Sleuth to fill in a column you define, on any dataset in Explore or Analyze.
- [Confidence scores and labels](confidence.md): How Sleuth scores every value, and what sourced, inferred, and unknown mean.
- [Check a dataset with Review mode](review-mode.md): See only the cells worth a second look, check their sources, and mark them reviewed.

*Updated November 9, 2026.*
