One of the most important benefits of document automation is the ability to read and process documents without human intervention. By automating routine tasks, companies can thus save time and money. However, in order for a document to be processed automatically, it must first be trained properly. This blog post discusses when a document is suitable for automation and how you can train your documents.

In Doc², you can see directly via the AI indicator how well the document is trained. The AI indicator is a measure of how well the OCR engine has been trained to recognise the document. The higher the score, the less likely it is that a human will need to intervene to ensure the document is processed correctly.

After each successful export, the AI indicator automatically increases. If someone intervenes in the document and changes the rules for the document, the AI indicator will be negatively affected.

For automatic document capture, the documents should be processed at least 4-5 times without any changes and the AI indicator should be at least 50%.

In the example we train a new invoice and export it 5 times with different values.

The first time the document is read in, the AI indicator is at the minimum level:

Let’s train the document and export it for the first time:

This way, the fields are read automatically when they are exported again and we get a recognition score that shows the read quality for the current document:

Furthermore, for each successful export, the AI indicator is influenced so that it will gradually move from the red to the green area in the graph.

Documentation: AI Indicators