The final extension is the Post-Final, where all the information about what the AI has understood, including possible human corrections, is transmitted, especially in the case of categorisation or extraction.

If the extraction is not categorised, it may consist of simple data, and information is also provided regarding whether the document has been discarded or not. When the KAPTO document reaches its final state, the extension utilises the information to correct the data presented in the mask. It indicates that the document has reached a completed form for this application.

Defining KAPTO entity fields involves specifying whether they should be identified by the AI or filled by another source. Another source can be an API that loads the documents or an extension that fills the fields.

In the categorisation model, KAPTO processes e-mail received from the customer's e-mail system. These e-mails contain the raw pages of the documents sent by the customer. KAPTO's task in the categorisation model is to understand the contents of the documents and split them up.

Once the e-mails have been categorised, KAPTO can apply the appropriate extraction model to each category, allowing for more efficient processing of the documents. The extraction models can identify specific fields within the documents, such as names, addresses, and account numbers, and extract that information for downstream processes.

Most of these fields are populated by the public API that loads KAPTO documents into the system. While working with categorisation and extraction, we can create fields for the model. However, in the categorisation model, we know that AI does not fill these fields by definition. Nevertheless, we can decide whether to show them to the user in the History or Verification phase or keep them as completely hidden fields used only for processing.

The highlighted fields are filled through a Post-AI process, which involves extracting information related to the document's claim.

The AI uses the information it has understood to query the legacy system of the insurance and locate the claim based on specific details such as the date, place, and notification data. The KAPTO extension uses this extracted information to identify the claim.

If multiple possible alternatives exist, but the AI is confident that the most probable one is correct, it moves forward and provides an option. However, if the Post-Evaluation extension cannot identify the claim confidently, it moves the documents into the Verification phase and requests human intervention.

The Post-Final extension examines documents that arrive in the Final Stage and searches for any other documents that belong to the same e-mail. It then consolidates the information to identify the claim accurately and ensure that all the documents in the same communication are positioned correctly.

‍