The latest addition to Google LLC’s public cloud is the Document AI Platform, which enables enterprises to extract information contained in digital and printed documents automatically using machine learning.
The new platform was announced today and is currently in preview.
At traditional enterprises, a lot of important information is contained in paperwork such as invoices. That information has to be extracted and loaded into back-office applications such as a firm’s accounting system so it can be processed in an organized way. The challenge is that data entry is still conducted manually in many cases, which takes up valuable time that workers could be spending elsewhere.
Google is promising to automate the process with the Document AI Platform. It brings together two of Google’s existing document processing services, Lending DocAI and Procurement DocAI, in a unified dashboard that workers can use to create automated data extraction workflows.
On launch, the platform includes more than 10 “processors,” pre-packaged data extraction engines each optimized to retrieve information from a different type of document. There are two general-purpose processors: one for regular documents and another for forms. They’re complemented by specialized engines for extracting data from financial documents like lending applications and procurement-related paperwork such as invoices.
All the processors “are created and fine-tuned to achieve industry-leading accuracy, helping businesses confidently unlock insights from documents with machine learning,” Google Cloud product manager Lewis Liu and product marketing manager Yang Liang wrote in a blog post. Over time, they elaborated, more processors will be added to cover additional document types.
Document AI Platform also includes an application programming interface for integrating with a company’s applications.
Besides freeing up employees for higher-value tasks, Google said, it can also help enterprises increase the accuracy of their business information by reducing errors related to manual data entry. The Alphabet Inc. subsidiary claims that one user, Belgian financial technology firm Unifiedpost, managed by increase data capture accuracy by 250%.
There’s a potentially big market for AI tools that can automate document processing because it’s a task that practically every company performs as part of its backoffice operations. Parsing sales orders alone represents a sizable automation opportunity. Conexiom estimates that some $8.4 trillion dollars of business-to-business sales orders are processed manually in the U.S. every year.
Google’s top rivals in the cloud market have also entered this segment. Amazon Web Services Inc. offers Textract for extracting information from documents, while Microsoft Corp. competes in the category with Form Recogniser.