Our Candis AI automatically detects document types, helping you manage your documents in Candis faster and more easily. In this article, you’ll learn how automatic document type detection works and what it means for your daily work.
What does automatic classification do?
As soon as a new document is uploaded to Candis, whether via email or drag & drop, it’s automatically checked for its type. The classification identifies what kind of document it is and assigns it to the appropriate document type. This saves time and reduces manual steps.
Note
Our AI-powered document classification is secured with strict data fencing. All data is processed exclusively within the EU and is not used for machine learning purposes, ensuring full compliance with financial and data protection regulations.
These document types are automatically detected:
Incoming invoices
Contracts
Offers
Delivery notes
Order confirmations
Documents that cannot be clearly assigned to one of these categories are automatically classified as invoices. The reason: Most documents in Candis are invoices - this ensures that documents are classified in a meaningful way when in doubt.
Where can I see which type was detected?
In the history of each document, you’ll find a note indicating that the document type was automatically classified. This gives you transparency about what the system has recognized.
Example:
“Document classified as invoice”
“Document classified as contract”
(If you don’t see this note, automatic classification hasn’t been activated for your account yet.)
What happens if the AI detects the wrong type? Can I change the document type?
In most cases, classification works reliably. If the document type isn’t recognized correctly, you can manually change it. Your correction helps us improve classification in the long run.
Good to know:
When you change the document type, most of the already captured data stays intact. In some cases, e.g. when switching from "Contract" to "Invoice", the data extraction might be re-executed to fill in the appropriate fields again. Usually, no information is lost in this process.
How does classification work in the background?
The classification is based on modern AI (Large Language Model). It analyzes the document content:
Text documents: If the document has digital text, it’s analyzed directly.
Scans or images: For image-based documents (e.g., scans), the first pages are processed by the AI for analysis.
Based on clearly defined criteria, the AI determines whether it’s an invoice, contract, or another document type.
What happens with e-invoices (e.g., XRechnung)?
E-invoices are also detected and processed accordingly. They go through the same classification process, with Candis recognizing that they are electronic invoices in XML format. These documents are then directly handled as invoices and processed further.