Help Center

Everything you need to get started and to run DokuScan day to day. The steps refer to the application at app.dokuscan.com.

Quick start
Templates and extraction modes
Uploading documents
Email inbox
Reviewing results
Export
Packages and scans
API and webhooks
Privacy and security
Common problems

Quick start

Sign up and log in. New accounts get 10 free scans, no payment method needed.
In the quick start, upload your own document or use the sample invoice.
Review the suggestion: DokuScan detects the extraction mode, fields and table columns. The analysis is free and uses no scans.
Save the template.
Upload more documents of the same type with this template.
Review the results and export them as Excel, CSV or JSON.

The right template is the most important step. A template fits one document type, for example the timesheets of one particular layout. Create a separate template for other document types.

Templates and extraction modes

A template describes which data is read from a document type. It consists of fields and an extraction mode.

Mode	Suited for
Single fields	Invoices and forms: one value per field and document.
Table/rows	Timesheets and delivery notes: each row becomes a record.
Both	Header data and rows, for example supplier and line items.
Image mode (Vision)	The AI reads the page images instead of the text. Useful for handwriting, stamps and difficult layouts. At most the first 10 pages are read.

Field types

Text: names, addresses, times and any other strings.
Number: amounts and hours as decimal numbers.
Date: formatted as DD.MM.YYYY.
Yes/No: checkboxes and confirmations.
Table: repeating rows inside a single-fields template, for example a list of line items.

For each field you can add a hint for the AI, for example where the value is located or how it is written. The technical key of a field is derived from the field name and appears in the API and webhook data. It must be unique per template.

Creating and maintaining templates

Create from a sample file: the assistant analyzes a sample document and suggests mode and fields. The analysis runs in the background. For multi-page files the assistant detects whether each page is its own document. If a suitable template already exists, it points it out. The trial run processes the document like a real scan.
Check and correct: test an existing template with a trial run. The assistant suggests corrections. Analysis and trial runs use no scans but are limited to 20 analysis steps per hour. Depending on the workflow, one analysis takes one to three steps.
Version history: every change is saved as a version and can be restored.
Default template: a template can be preselected for uploads.
Duplicate: copy an existing template and adapt it for similar documents.

Uploading documents

Formats: PDF, JPG, PNG, WEBP, TIFF, BMP and HEIC, up to 20 MB per file and up to 25 files at once.
Multi-page PDFs: with the setting "Split multi-page PDFs" each page is processed as its own scan, e.g. one receipt or timesheet per page. Without it the whole file counts as one document. If DokuScan detects a multi-page PDF on submit without the setting, it asks whether to split or process it as one document.
Background processing: the browser does not need to stay open. Progress is shown.
Duplicate protection: the same file is not processed or charged twice with the same template.
Credit on failure: if a scan fails it is credited back automatically.
Limit: up to 200 scans per hour.

Email inbox

Each template has its own inbox address. You find it in the template under "Email inbox". Documents sent as attachments to this address are processed automatically with that template, without logging in.

Supported attachments: PDF, JPG, PNG, WEBP, TIFF, BMP and HEIC.
Each attachment is processed like an upload and uses one scan.
The email body is not stored. The sender address is saved as a note on the scan.
If there is a problem (no valid attachment, quota used up, unknown address) the sender receives a reply by email.
The same limits as for uploads apply.

Treat the inbox address like a password: anyone who knows it can trigger scans.

Reviewing results

Review view: the original and the extracted values side by side. Rows can be corrected, added and deleted. Changes are saved when you move on.
Rotate: images can be rotated left or right in the review view. The rotation only affects the display.
Confidence: each scan shows how sure the AI was. From 80 a value counts as high, from 50 as medium, below that as low. You should check low values.
Validation rules: per field you can set min and max values, allowed options and a check for dates in the future. Violations are flagged and the scan gets the status "Validation failed". The data is still available.
Partial result: if the model response or the document text was truncated, the result is marked as partial. Entries may be missing at the end.
Learning from corrections: values you save in the review (e.g. employee names or work sites) are remembered by the template and passed to the AI on the next scan as a reading aid. Corrected misreadings are remembered too. Only short, recurring text values are learned, no numbers, dates or free text. In the template you can switch learning off, overall or per field, and reset the learned list. The model itself is not trained, and results still need to be reviewed.

Export

Formats: Excel, CSV and JSON. For CSV you can choose semicolon, comma, tab or pipe as the separator.
You can export single scans from the review view or several scans together.
Export mappings: rename, split and merge columns and convert types. The menu entry appears after the first scan.
Export configuration from a sample file: upload a sample file of your target system (CSV, XLSX or JSON). DokuScan suggests which values belong in which column and shows a trial run with your latest scans. The configuration can then be adjusted in the editor. One header row with one record per row is supported, formats with several record types are not. The assistant is limited to 10 analyses per hour.
In the Excel and CSV export, Yes/No fields appear as "yes" and "no".

Packages and scans

One scan equals one processed document. Each page of a PDF counts as one scan, whether or not it is split.
Packages are bought once, there is no subscription. Scans do not expire.
Further packages are added to your existing balance.
New accounts have 10 free scans. The analysis for creating a template uses no scans.
For every purchase we issue an invoice and send it by email.

View pricing

API and webhooks

Every purchased scan package includes a REST API with webhooks. You manage API keys and webhooks in the menu under "API & Webhooks". The interface is described in the API documentation.

API documentation

Privacy and security

Documents are stored encrypted in the EU (Frankfurt).
Your documents and data are not used to train AI models.
Accounts are strictly separated from each other.
We provide a data processing agreement (DPA) on request.

The full details are in the privacy policy.

Common problems

The result is wrong or empty.

First check that the right template and extraction mode are selected. A template fits only one document type. Test an existing template with "Check and correct". For handwriting or difficult layouts, image mode helps.

A scan has the status "Failed".

The scan is credited back automatically. Upload the document again. If the error persists, use "Get help".

A document was skipped.

The same file has already been processed with the same template (duplicate protection). With a different template a new scan is possible.

The status is "Validation failed".

At least one validation rule was violated. The data is available. Check and correct the flagged values in the review view.

The upload is rejected.

Check format and size (PDF, JPG, PNG, WEBP, TIFF, BMP, HEIC, up to 20 MB) and whether your quota is used up. New accounts have 10 free scans, after that a package is required.

Only the first page of my PDF was captured.

If a PDF contains several documents (e.g. one timesheet per page), enable "Split multi-page PDFs" when uploading. Otherwise the whole file counts as one document. Move an already scanned file to the trash first, otherwise duplicate protection applies.

Names or places are read incorrectly.

Correct the values in the review and save. The template remembers them and uses them as a reading aid in later scans. You can also store a fixed list under "Allowed values" of a field.