Before we dig deeper into Document AI, let’s first understand what Document Capture Solutions are. I started my career with these Capture Solutions. I was amazed at the capabilities of this advanced technology. Capture solutions extract the textual data from documents and images using OCR engines, and then convert them into actual text. Large enterprises use capture solutions to manage big volumes of documents in various formats to ensure an effective document management system is in place.
The providers for enterprise capture solutions are big leaders such as IBM, OpenText, Kofax, and many others. There are new solutions coming up from Google, Amazon, Rossum, and Microsoft that are taking advantage of cutting-edge technologies not just to provide OCR, but also to offer intelligent predictions. The main advantage of these technologies is that one can pick and choose parts of solutions applicable to their use cases.
We have partnered with Google to explore its new solution – Document Understanding AI, and to integrate it with other enterprise tools to build an end-to-end document management system. Currently, we are implementing a solution for one of our clients to automate their utility bills processing by integrating Google Document AI with ServiceNow and Box. We will dive into the details of this solution in the coming series of blogs.
What is Google DUAI/Document AI?
Google Document Understanding AI or Document AI is a new cloud based capturing solution provided as an API service by Google. Launched in 2019, DUAI is a simple solution based on machine learning technologies that has the ability to process a great variety of unstructured data.
DUAI provides parsing processes such as form parsing, table parsing, and invoice parsing. The type of the document needs to be defined before it is processed based on the kind of data one wants. The following are the two options that one can select from:
a) General: It is used mainly for form and table parsing. It provides results as key-value pairs, tables, and text that is generated by OCR.
b) Invoice: This option provides data that can be matched with predefined annotations.
Why does Document AI stand out?
In contrast with many of the existing standard solutions, Document AI stands out as a modern solution by taking advantage of all the new age tools available today. It is a cloud solution that helps avoid all the hustle to setup multiple servers, specialized components, and configurations for individual products. It uses Machine Learning, NLP, knowledge-based graphs, and other similar technologies to eliminate the tedious processes of defining and testing templates or looking for keywords. All sets of documents work without failing the process or application. Moreover, there is no specialized training needed to work with Document AI as it is very user-friendly and intuitive to use.
API Support
Document AI offers client support in multiple programming languages. The REST services that act as a middleware enable Document AI to be integrated with any solution at any stage of the process. This also enables the user to either use Out of the Box provided or self-personalized verification screens.
High Accuracy
Google’s Document AI solution provides 96% data accuracy. Check out the sample below to understand its caliber.
Limitations
As with any product, Document AI also has a few limitations. One of such limitations could be capturing complex line items. For instance, if an invoice has line items occupying more than a single line, Document AI fails to capture those lines completely.
Conclusion
All the discussed features and functionalities of Google AI provide users with a great liberty to use any combination of solutions, integrate with several other tools, enhance their existing solutions, and create their own solution with their own choice of software. Are you interested in leveraging Google Document AI to automate your manual data entry processes? Engage with us today!
