Google finds new Document Understanding

Among a range of new artificial intelligence (AI) tools announced by Google this week is Document Understanding AI, promoted as a way to automatically classify, extract, and structure data within scanned documents

Currently in Beta version, the tool is designed to help automate document processing workflows via a scalable, serverless platform.

Google Cloud Group Product Manager Levent Besik in a blog post, said “customers that use custom document classification have achieved up to 96% accuracy. Document Understanding AI easily integrates with technology stacks from partners and third parties—Iron Mountain, Box, DocuSign, Egnyte, Taulia, UiPath, and Accenture are already using it today

Jim O’Dorisio, Senior Vice President for Emerging Commercial Solutions, Iron Mountain, said "As the world's leading information management provider, Iron Mountain scans over 627 million pages every year as part of our digital transformation solutions. Google Cloud's Document Understanding AI helps us identify form fields, text passage, tables and graphs, as well as customer-specific keyword matching, for customized workloads,” says

“Document Understanding AI provides a foundation to help us deliver a far more valuable set of services to our customers—assisting them in automated data understanding, enabling compliance, business value, and delivering peace of mind."

Andrew Moore, head of Google Cloud AI, told the Wall Street Journal, “The CIO is the critical person in this. We’re focused on specifically transforming companies’ digital presence, and we’re really determined to go [after] central business problems.”

In a related announcement, Google revealed that Contact Center AI, is now in beta, along with Virtual Agent, Agent Assist, and Topic Modeler. Contact Center AI builds on Google’s Dialogflow Enterprise Edition and interacts with customers over the phone, fielding incoming calls and using natural language processing to suggest solutions to common problems. If the virtual agent can’t solve an issue, it hands the caller off to a human agent and presents the agent with information relevant to the call at hand.