Introducing the AI-Powered Compliance Documents Recognition

Last updated November 01, 2023
Written by Joel Martin


One of Felix's goals for clients and their vendors is to streamline onboarding for speed and to simplify compliance procedures. We recognised that filling in compliance document details can be a time-consuming challenge for vendors, and organisations often have to engage with vendors to rectify inaccuracies, such as policy numbers or provider names which can lead to delays in the onboarding process and friction between companies.


To make progress towards this goal, we have dedicated resources to develop an innovative AI-powered technology that can automatically identify compliance documents and extract essential information. It can expedite vendor onboarding, reducing the time spent on manual data entry. Additionally, this technology can enhance the accuracy of initial document submissions by vendors, reducing the need for organisations to follow up on corrections. This improvement in efficiency can lead to stronger vendor relationships through a smoother onboarding process with reduced friction.

How does it work?

Vendors can now benefit from automatic document scanning when they upload specific compliance documents. The uploaded files will undergo scanning to extract key metadata using AI-powered technology, which is designed to comprehend and extract data effectively – specifically for common compliance documents. 

After the extraction process is finished, the relevant data will be automatically populated into the appropriate fields in Felix, eliminating the need for vendors to manually fill out the form.

Vendors retain the option to review the populated data and make corrections if any errors are found.

How accurate is the data extraction?

The AI model is proficient at extracting crucial information and has been trained specifically on the various types of compliance documents vendors typically upload in Felix. 

Nevertheless, certain factors like the quality of the uploaded documents (e.g. low resolution or poor scan quality) may affect the accuracy of processing. 

If the model is not confident that it has recognised certain fields, they will be left empty for the vendor to populate manually.

Compliance document types compatible with AI-powered document recognition

The following compliance document types were used to train the AI model and hence are currently the only type of documents we attempt to process and extract data from:

  • Public Liability Insurance
  • Product Liability Insurance
  • Professional Indemnity Insurance
  • Public & Products Liability Insurance (combined policy)
  • Contract Works Insurance
  • Motor Vehicle Insurance
  • Plant & Equipment Insurance
  • WorkCover/Workers Compensation Insurance

Felix will continue to monitor for opportunities to expand the coverage of this innovation. We welcome your feedback on where it could help you and your business.

Was this article helpful?