Using the generic connector for AI - Process diagrams to support AI services

The following diagrams explain the processes that have been implemented in the HighQ AI Hub in order to support connection to AI services using the generic connector for AI. These should assist you to:

• Understand the end-to-end process
• Design supporting APIs that will support the process

Adding a new AI service to the HighQ system

This process is initiated by the HighQ System Administrator.

It is assumed that the service is configured, and a client ID and token have been generated by the service and sent to the HighQ System Administrator (for example by secure message).

Using the generic connector service in a site

The AI service must have an existing project ready to receive files for the HighQ AI Hub to use the service.

This project roughly corresponds to a HighQ site, and contains files that are to be processed in the AI service.

It is assumed that the project will have a specific extraction ‘template’ assigned, such that files will all have the same data extracted. In future, HighQ may wish to offer users of the HighQ AI Hub the ability to create a project on your AI service, and select a specific template to apply (if more than one extraction template is available on your AI service).

For example, your AI service may offer a template for extracting commercial contract details, and a different template to extract employment contract details.

If a specific project is created for each template, then the HighQ AI Hub can send appropriate files to each project.

Processing files and returning extracted data

This process is initiated by the AI service completing the extraction process.

The service will call the AI Hub to notify that a specific file (or ‘job’) has completed. This will prompt the AI Hub to pull the collection of answers to each question from the service.
The pattern used here is that the extraction service creates a collection of question/answer pairs.