This page defines the terms used across the documentation. Each definition here is the authoritative one for the site; the link in each entry points to the page that explains the term in full.

Batch

A batch is the group of files created by a single upload, and it is the unit you process, review, and export. Every batch gets an automatically generated name, and how many files it can hold depends on the account's plan. See Batches.

Column

A column is one field you want extracted, and each column becomes a column of the output table. A column has a column name and, optionally, a Format Requirement, and a template is a saved list of columns. See Templates & Columns.

Computed column

A computed column gets its value by calculation instead of by reading a value that is printed — for example a line total the document does not show. The calculation is written in the column's instruction, in its name or its Format Requirement, and it runs in the same pass as the rest of the extraction. See Custom Columns; for the switch that asks the model to reason more about a calculation, see Precision+.

Conflict handling

Conflict handling is the template-wide choice for what happens when two or more files in one merge group report different non-blank values for the same column: combine them into one cell, keep every value as its own row, or keep the first or last value. It applies to the whole template, not to individual columns. See Multi-Page Merge.

Credits

Credits are the unit the product charges for processing: a run is charged per processed page at the account's Model Tier rate. Standard is the only tier open to accounts today and costs 1 credit per page. Some parts of the interface still label this unit Points; the two terms refer to the same thing. See Model Tier.

Custom column

A custom column is any column whose value the AI produces for you, whether it is read directly from the document, calculated from what the document shows, or inferred from context. All three kinds are set up the same way, with no separate mode to switch on. See Custom Columns.

Direct extraction

Direct extraction reads a value that is printed on the document and returns it as it appears. It is the default: unless the instruction asks for a calculation or an inference, the AI reads the value rather than computing or rewriting it. See Custom Columns.

Document (File)

A document is a single item inside a batch: one uploaded image, or one page of a PDF, Word, or text file. The web app labels these Files, while the API calls the same unit a Document. When Multi-Page Merge refers to "the same document", it means a real-world document that arrived as several of these files. See Batches.

Format Requirement

A Format Requirement is an optional instruction attached to a single column that controls how its value should come back — for example converting an amount to a plain number or normalizing a date — and it is also where a calculation or a decision rule is written. It applies only to that column; leaving it empty returns the value as it appears. See Templates & Columns.

Inferred column

An inferred column asks the AI to decide something that is not written on the document, using the document as a whole — for example placing a receipt into a Category. You describe how to decide, including the allowed answers when you want the result constrained. See Custom Columns.

Merge group

A merge group is one set of results that Multi-Page Merge treats as a single record: the rows from the several files that belong to the same document. Which files form a group is decided by the merge mode chosen in the template settings. See Multi-Page Merge.

Model Tier

A Model Tier is the run configuration an account processes on. It selects the models used and how many credits each processed page costs; Standard is the default and the only tier open to accounts today, at 1 credit per page. A tier is assigned to an account by the ImageToTable.ai team, and the tier is fixed for a file the moment processing starts. See Model Tier.

Multi-Page Merge

Multi-Page Merge combines the extracted results that belong to the same document into one record when that document was uploaded as several files. It is configured in a template, works on rows within a single batch, and takes effect when you export. See Multi-Page Merge.

Precision+

Precision+ is a switch on the dashboard that asks the model to spend more time reasoning before it answers. It changes how much the same model reasons without changing the account's Model Tier or the credits charged per processed file. See Precision+.

To Table

To Table is the processing mode that extracts the columns of a template from each file and returns the results as rows of a table, exported as an Excel workbook. It is one of the two modes in the Select Mode switch; the other is To Word. See Templates & Columns.

To Word

To Word is the processing mode that returns each file as a whole page instead of extracting columns, and exports as an editable Word document. It needs no template, keeps the page's headings, tables, alignment, colors and pictures, and rebuilds flowcharts and data charts as editable shapes and native Word charts. See To Word.

Template

A template is a saved, reusable list of the columns you want extracted, so the same set of columns can be applied to batch after batch without retyping it. An account can hold up to 30 templates, and a template also carries settings that apply to the whole list, such as Multi-Page Merge. See Templates & Columns.