Jump to Content
Public Sector

Automate document processing with Document AI for government

October 19, 2022
https://storage.googleapis.com/gweb-cloudblog-publish/images/Hero_Image_1.max-2600x2600.png
Rajat Gupta

Product Manager, Google Cloud

Jonathan Chen

Solutions Architect, Google Cloud

The Challenge 

Many government agencies remained overwhelmed with paper, even as the pandemic drove digitization and modernization. As a result, government agencies continue to look for better ways to streamline manual processes that overwhelm their workers and take focus from higher value initiatives. Modernization helps agencies better serve citizens, improve benefit delivery speed, and resolve technical debt accumulated in legacy systems, but today’s fragmented toolset does not meet governments where they are given their staffing level, skillset, and budget.

Agency leaders can use Document AI to help process paper and images at scale to better deliver services to their constituents. Document AI helps accelerate the speed of innovation, cuts document and image processing   time, and extracts information at scale by leveraging Artificial Intelligence (AI) to ensure high accuracy. The newly announced Document AI Workbench helps ingest, classify and extract relevant information from custom documents. Combined with the built-in Human in the Loop feature, Document AI for government provides the human review safety net to ensure information is extracted with a high level of accuracy. The custom machine learning (ML) models can be consumed by applications that are hosted in other public clouds, robotic process automation (RPA) vendors like Appian, UiPath, ServiceNow and on-premises services to integrate with existing or new systems.

Introducing document intake accelerator

Document intake accelerator is an open-sourced, pre-packaged and customizable solution. It incorporates a mature end-to-end workflow for document processing and management to speed implementation and reduce costs for federal, state, and local governments and academic institutions. Document intake accelerator creates a quick adoption path to modern document processing technology utilizing Google AI. It helps developers bootstrap the baseline document intake pipeline while applying Google’s best practices like Kubernetes-based microservices, cloud security frameworks and highly scalable architecture.

https://storage.googleapis.com/gweb-cloudblog-publish/images/Diagram-1.max-800x800.png
Automated document processing steps in Document Intake Accelerator

Document intake accelerator automates a document intake process through the following steps:

  • Each document is automatically classified by Vertex AI to a particular document type.

  • The Extraction module uses Document AI parsers to extract information from a document. 

  • The Validation module and Matching module evaluate each supporting document for its authenticity and eligibility.

  • Lastly, each document is sent to Human in the Loop staff as the final checkpoint to review, approve or reject this document. 

https://storage.googleapis.com/gweb-cloudblog-publish/images/Diagram-2.max-1000x1000.png
Document AI for government parsers with high accuracy in entity extraction.

Document Intake Accelerator reduces development time with highly configurable modules and workflows to address a wide range of scenarios and document types. In addition, Document Intake Accelerator provides API endpoints to enable easy integration with third-party or custom-built systems to enable end-to-end document lifecycle automation.

https://storage.googleapis.com/gweb-cloudblog-publish/images/Diagram-3.max-1100x1100.png
Document Intake Accelerator provides API endpoints to interoperate with the existing systems.

Document AI applicability across functions

Since its launch in 2020, Document AI has seen wide adoption by government agencies including the State of Hawaii. The State of Hawaii Safe Travels program leveraged Document AI to extract travel and health information for all visitors and helped the state safely reopen to tourism.

We see great potential for leveraging Document AI across the government. Some of the use cases where we can make a difference are: 

  • Procurement and supply chain system: government agencies have responsibility to manage huge supply chain systems that are responsible for tracking and moving goods. We can use our pre-build ML models like invoices, Contracts, BIll of Lading to effortlessly extract information out and accelerate the system.  

  • Processing tax documents: Document AI has out of the box ML models that can read information from various tax forms like W2, W9, 1040, 1099, 4506, etc. These forms are a huge part of various benefits programs and by using them the government can realize instant benefits by accelerating the data extraction. 

  • Academic institutions: universities can use Document AI for extracting transcripts and  resumes to support students to align with their learning goals and job searching. 

Conclusion

Document AI brings a much needed toolset to help governments create custom document ML models at scale, achieve high accuracy with human reviews, and provide a solution accelerator that enables results in weeks - not months. This highly customizable platform gives governments and agencies the power to tackle document processing challenges at scale with lower costs and higher efficiencies. To get started, try using Document AI and enable the workflow using the Document Intake Accelerator

Learn more about how the government is accelerating innovation at our Google Government Summit, taking place in person on Nov. 15 in Washington D.C. Join agency leaders as they explore best practices and share industry insights on using cloud technology to meet their missions. Can’t join in person? Register for the Google Government Summit keynote livestream session. 


Posted in