This quick start guide will walk you through the Rossum app and show you all the basics. You will learn how to:
- Create an account
- Upload your documents
- Validate the data
- Export the data
- How to customize data capture
Create an account
To create an account, head to the Rossum registration page.
In the first step, enter your personal information. That includes First name, Last name, Company name, Phone number, and Business email address you want to associate with the account. Then click on the “Get Trial Access” button to continue.
Next, you will see a new window where you need to create a secure password. Once ready, click the “Get started!” button to open your Rossum account.
Please note your trial account is free for <300 documents per month and is fully configurable.
When you log in to your Rossum account for the first time, you will see the document type options. Choose the one you want to process in the Rossum app.
Note: Although Rossum currently supports only pre-trained data fields for processing invoice and purchase orders, the technology is document agnostic and can extract data from any structured document, including receipts, shipping documents, claims, packing lists, etc.
And as the last step, you need to select your region to define the list of pre-trained fields that Rossum will extract. Don’t worry, you can change it later.
Once you select it, you will be able to upload your first document and process it in the Rossum app:
Note: Automated learning of your documents is available only in higher editions.
Welcome to the app
After logging in, you will see the main screen.
On the left side, you will find a list of Queues. Each queue is an organizational unit that typically represents a specific document type that needs to be processed. Using the fully customizable left panel, you can create, remove, or group queues as needed.
Read more: How to configure fields for data extraction.
In the center of the screen, you can see a list of your documents or samples we provided.
Upload your documents
Start uploading your documents by clicking the “Import” button or sending them to the provided email address (you can also do this via the API).
You can import PDF, PNG, JPEG, TIFF, XLSX, and DOCX files or scanned images of your documents.
When you click “Done“, the AI will automatically begin extracting the data fields. The AI extraction typically takes between 30-60 seconds.
After processing, you will find the documents in the “To review” tab in case human review is required.
Watch this short Product Walkthrough video to see Rossum in action:
Validate the data
A grey checkmark means that an AI validated the field, while a green checkmark indicates that a human validated the field. Start with the fields that don’t have check marks if you want to focus on the most problematic ones.
To check the fields one by one, press Tab or Enter to skip the green and grey checkmark fields. Please remember that you must review and correct the red-flagged fields. You will not be able to export the data otherwise.
If you notice that Rossum did not extract a data field correctly, you can make adjustments by:
- Clicking and pointing
- Dragging the rectangle
- Typing the text directly
Export the data
After the validation, click on “Confirm“. The document will appear either in the Confirmed tab (you can enable/disable this view in the Queue’s Basic settings) or in the Exported tab on the main dashboard.
There you can choose to download the data in different formats (XLSX, CSV, XML, JSON, via API)
Customize the data capture
Once you are confident with using Rossum, you can customize the captured Data fields. Read more about the out-of-the-box extracted fields and how to set up a custom field.
Note: Rossum returns confidence scores for missing fields for the header fields returned by the Generic Engine. It is worth noting that even after this change, the logic of the automation settings remains the same, so you should see no difference in the system’s behavior. Please contact email@example.com if you want to automate missing fields based on confidence thresholds.