A template is a saved, reusable list of the columns you want extracted from a document. Once you save one, every future batch can reuse it instead of describing the same columns again.
What a template is
A template holds the list of columns that get extracted. Each column becomes a column in the resulting table. The template itself is just the saved container for that list, so the same set of columns can be applied to batch after batch without retyping it.
This is the difference between a template and a one-off run:
- Saved template — a named, reusable column set. It stays in your account and can be switched on any batch, or reused through the other entry points listed further down this page.
- One-off run — the columns exist only for that single run and are not saved for reuse. On the web app this is what happens when the selected template has no columns defined: the AI infers the table structure from the document itself. Through the API, you can also pass a column list for one request only, without creating a template first.
Creating, switching, and editing templates
Templates are managed from the Data Columns panel on the dashboard (To Table mode). The panel header shows the currently selected template and three actions:
| Action | What it does |
|---|---|
| New | Opens the New Template dialog. Give the template a name, and optionally base it on an existing template to copy that template's columns. You can also browse the Preset Library to start from a ready-made set. |
| Switch | Opens the Switch Template dialog, which lists your templates. Each row offers Select, Settings, Share, and Delete. |
| Settings | Opens Template Settings for the current template — see Template-wide settings below. |
To add a column, type its name into the box under the template title and click Add Column. If no template is selected yet, one is created for you automatically. To change a column, click its edit icon in the list; to remove it, click the delete icon.
What a column is made of
Each column has exactly two parts: its column name and, optionally, a Format Requirement. Both are sent together when the document is processed.
| Part | What it controls |
|---|---|
| Column name | The label of the output column, and the instruction for what to find in the document — for example, Invoice Number or Total Amount. The name is also what the AI looks for, so it should describe the data you want. |
| Format Requirement (optional) | An extra instruction for how that column's value should come back — for example, converting an amount to a plain number, stripping a currency symbol, or normalizing a date to a fixed format. It applies to this column only. If you leave it empty, the value is returned as it appears in the document. |
If a column has a Format Requirement, an information icon appears next to its name in the list so you can read the instruction without opening the edit dialog. A column with no Format Requirement simply extracts the value directly.
The Custom Columns page explains the different ways a column's value can be produced — read directly from the document, calculated from other values, or inferred from context.
Changing column order
Columns are extracted and returned in the order they appear in the template. To change that order, drag a column by its drag handle in the list. The new order is saved immediately, and the output table follows it. This is useful when you want the resulting spreadsheet to lead with a particular column.
Template-wide settings
The Template Settings dialog (opened with the Settings action on the Data Columns panel, or per-row in the Switch Template dialog) holds settings that apply to the whole template rather than a single column:
- Global Format Requirement — one instruction that applies to every column in the template, on top of each column's own Format Requirement. Use it when the same formatting rule (for example, a date style or a number format) applies across the board.
- Merge Multiple Results — combines several extracted rows into one. This is the same feature described on the Multi-Page Merge page.
AI-Suggested Columns
When you start processing a To Table batch whose selected template has no columns yet, and the batch contains at least two files, the web app can propose a column list for you. A few of the files are sampled, the AI reads them, and the AI-Suggested Columns dialog opens with its suggestions.
- Review the suggested column names, edit any of them, add your own, or delete ones you do not want.
- Click Apply & Process to save the list into the template and start processing.
- If the suggestion fails, choose Retry or Define manually.
The suggestion is skipped when the template already has columns, or when the batch has only one file — in those cases processing starts directly.
How many templates you can have
An account can hold up to 30 templates. When you try to create a new template after reaching that limit, the app shows this message and the new template is not created:
You've reached the limit of 30 templates. Delete an existing one first.
To free up a slot, open the Switch Template dialog and delete a template you no longer need.
Where templates can be reused
A template is not tied to one batch. The same saved column set can be applied from any of these entry points:
- Web app — select the template in the Data Columns panel and process any batch with it.
- Google Sheets add-on — the sidebar uses the same saved templates, so a template built on the website is available there as well.
- API — pass the template's identifier when starting a batch, instead of listing columns in the request. See the API Templates & Fields reference for the developer-facing details.
Templates can also be shared as text: the Share action in the Switch Template dialog produces a block you can send to someone else, who pastes it into the column-name box and clicks Add Column to import the columns and settings into their own account.