Ava AI – Models

This overview shows the available standard AI models for use in Ava AI. Refer to as AI models we have a collection of algorithms and mathematical structures that are trained to specific data has been trained to perform specific tasks. These AI models can occur in various forms including neural networks, decision trees or vector machines. They are used to recognize patterns in data, make predictions, identify and extract data or perform other tasks, depending on the specific application requirements. See the following official models per API. In fact there’s a lot more. Please contact us.

Ava AI Extract API

ava-extract-policy-building-generic-1
This AI model was trained on building insurance. It is a standard model for extracting data from German-language insurance policies for residential and commercial buildings from various providers. This model can identify and extract the following data from insurance policies, among others: customer, agent, and insurer data, general contract details, as well as coverage scope data including premiums and insured buildings.

ava-extract-policy-business-generic-1
This AI model was trained on commercial insurance. It is a standard model for extracting data from German-language commercial insurance policies from various providers. This model can identify and extract the following data from insurance policies, among others: customer, agent and insurer data, general contract details, as well as coverage scope data including premiums, locations of insurance, building information, and types of businesses or professional activities including revenues.

Ava AI OCR API

The following AI models described below are suitable for recognizing and extracting text from files and can be applied in accordance with the service Ava AI OCR. The following standard models are available:

ava-ocr-document-transform-generic-1
This AI model is a standard model for extracting text from files encoded in UTF-8, whose content is in German language. This model extracts text while preserving readability.