Understanding the Tabular view
The Tabular view is one of three views in a Collection. It shows structured data extracted from documents in a table layout.
Columns represent fields extracted from documents (e.g., Contract name, Signing date, Key clauses)
Rows represent individual documents or data entries
Open the Tabular View
Click a Collection from the left sidebar to open the Collection page.
At the top, you’ll see the Collection name, with tabs below it.
Switch to the Main tab to open the default Tabular View.
At the top of the Tabular View, you’ll see two main editing actions:
“Add a row” and “Add a column”.
Add a column
Click “Add a column”.
A dropdown menu will appear with AI Autofill enabled by default.
Enter:
Column name (e.g. Key clauses)
Select the Type
A short column description to help the Agent understand what to extract
(e.g. Clauses related to shipping, returns, or breaches that may affect company revenue).
With AI Autofill on, the Agent will automatically generate the full instruction for the column and populate it once the documents are processed.
If you turn AI Autofill off, you’ll need to manually define the column type and instructions yourself.
Knowledge: The role of the first column
The first column has a special role.
It defines the primary entity the Agent works from, such as:
Uploaded files (e.g. contracts, reports)
Company names
All other columns are generated based on this first column, such as:
Parsed results from uploaded documents
Information retrieved via web search for a company
Add a row
Click “Add a row”.
Upload a file or files (single upload or batch upload is supported).
Once the first column of new row(s) is filled, AI Autofill columns are generated automatically, while non-AI Autofill columns require manual input.
Examples: Building a Collection for Company Operating Reports
Scenario
In many companies, operating reports are created across different time periods and regions, often in different formats.
The goal is to create a Collection that brings these reports together, structures their content, and turns them into a knowledge base for generating future insights.
Step-by-step
Define the structure with the Agent Start by creating a blank Collection, then open the Chat panel on the right to talk with the Agent. Ask something like:
Please structure the operating reports I upload into the following fields: **Report date**, **Summary**, **Operational risks**, **Operational strengths**, and **Actionable improvement points**.
The Agent will use this instruction to define the initial table structure.
Switch to the Tabular View to review the column names generated by the Agent based on your request.
Click “Add a column” in the top-right corner:
Setup:Column name: Supplementary cases
Type: Text
AI Autofill: Enabled
Column description: Extract the key themes of this reporting period from the report, search for relevant cases online, and summarize the cases in bullet points.
Click “Next” to continue.
The Agent will automatically generate the full instruction for this column and enable the Web Search tool based on your description. If you open Context, you can also see which columns or sources the Agent references when generating content for each column.
Click “Add a row”, then select “Upload a file” in the new row.
Upload one or multiple operating reports into the first column.
Once the file is uploaded into the first column, the Agent will automatically generate content for the other columns based on their definitions.
Result
This example allows you to combine reports from different formats into a single, structured Collection that’s ready for analysis and insights.