Skip to main content
Collection data table
The Data tab of a collection is a spreadsheet-like table for exploring and managing records. It is designed for large datasets: rows are paginated and loaded on demand, so collections with thousands of records stay fast.

​
Reading the Table

Each column header shows the column name and a type badge (text, number, date, boolean, url). Values are formatted according to their type: Structured values (nested objects) are displayed as JSON.

​
Sorting

Click a column header to sort ascending; click again to sort descending. The active sort is shown with an arrow (↑ or ↓) in the header. Sorting applies to the whole collection, not just the visible page. Pagination controls at the bottom of the table let you:
  • Move between pages with the ◀ ▶ arrows or by clicking a page number
  • Change the page size with the rows per page dropdown (10 to 50 rows, default 25)
  • See your position with the “Showing X–Y of Z” counter

​
Selecting and Deleting Records

1

Select rows

Use the checkbox on each row, or the header checkbox to select the whole page. A badge shows how many rows are selected.
2

Delete

Click Delete in the selection bar. The selected records are removed and the table refreshes.
Deleting records is immediate and permanent. If agents rely on this collection, make sure the records are no longer needed.

​
Data Written by Agents

Collections linked to agents evolve outside the UI: agents insert and update records during conversations. Two things to keep in mind:
  • New columns can appear. Schemas are flexible — if an agent starts recording a new field, it shows up as a new column.
  • System fields are protected. Record identifiers, creation timestamps, and ownership metadata are managed by the platform and cannot be modified, even by agents.
To audit what agents are doing with your data, every insert, update, and deletion is recorded as an audit event — see Using Collections with Agents.

​
Best Practices

Sorting by a suspect column is the quickest way to find outliers — empty values group together, and malformed entries stand out at the top or bottom of the ordering.
Reviewing data quality? Use 50 rows per page to scan quickly. Comparing individual records? Drop to 10 for readability.
Select and delete rows page by page rather than removing large swaths at once — it is easier to verify what you removed.

​
Next Steps

Visualizations

Turn your records into charts

API Reference

Query and update records programmatically