Apply AI
to Extract Data from
Non-structured Documents

The solution leveraged several open source libraries for AI and Machine Learning, along with AI services from Azure. The system uses Python for the machine learning, Natural Language processing and AI functions, and uses C# for the integration of the Python based services into the production environment. The solution is hosted in the Azure cloud and is fully elastic in its deployment. When the system is not in use the footprint is very small. When demand occurs and grows the automated capacity management algorithm, deploys servers, and other resources to meet demand, and then scales down when demand drops.

The value proposition of using this tool is three fold:

1. The system learns and adapts to changes in the type and layout of the documents that it receives.
2. The system scales up or down depending on demand.
3. The system processes documents efficiently and effectively freeing up credit analysts to focus on issue resolution and provide a consultative approach to credit approval, rather then being focused on routine low value add processes.