Google LLC’s artificial intelligence tools are already considered to be some of the best in the business, but businesses are always looking for newer and easier ways to apply AI to their applications.

With that in mind, Google Cloud today is adding more AI services to its portfolio in addition to a bunch of new tools that should enable enterprises to build their own AI-based applications more easily. In a wave of AI announcements at Google’s Cloud Next conference today, the public cloud infrastructure giant announced several new products that give enterprises a way to use AI to address common business challenges.

The new services include beta availability of Document Understanding AI, which works by scanning documents and transforming them into structured data so as to help automate document processing workflows.

“This means you can take advantage of the facts, insights, relationships and knowledge hidden in your unstructured documents and start making data-driven business decisions faster and more accurately,” Levent Besik, a group product manager at Google Cloud, wrote in a blog post.

Meanwhile, Google’s Contact Center AI service, which uses AI to assist with customer service requests, is now available in beta too. Released as an alpha service last year, it enables companies to build things such as virtual agents. It also provides Agent Assist and Topic Modeler capabilities, so human customer service agents can quickly find the information they need to help the customers they’re dealing with.

Contact Center AI gains more partners as well, with the likes of unified communications firm Avaya Inc., plus Accenture Plc. and Salesforce.com Inc. all joining the party. The idea is to integrate Contact Center AI with these companies’ platforms so their users can also benefit from the assistance it provides.

Google has also updated its Cloud for Retail services, which provide a range of tools to help retailers benefit from AI. New tools in the box include Vision Product Search, which is now generally available and enables retailers to build applications that integrate visual search functionality.

So, for example, a customer can simply photograph an item and use that to find similar products in a retailer’s catalog. There’s also a new Recommendations AI tool in beta that retailers can use to deliver personalized recommendations to their customers. Finally, AutoML Tables, also in beta, can be used by retailers to create machine learning models that can predict things such as future sales, thereby helping them maximize revenue and optimize product portfolios.

All that demonstrates how Google is looking to target industry verticals as a way of making AI available to more users, said analyst Holger Mueller of Constellation Research Inc.

“Google now goes into horizontal and vertical use cases,” Mueller said. “Document Understanding AI and Contact Center AI are key ingredients for CTOs to build next-generation apps in the area of automation. Along with [Google Cloud CEO] Thomas Kurian’s newly announced strategy of making Google Cloud more vertical, this is where Google Cloud for Retail fits in.”

In a broader sense, the AI Platform has to potential simply to make innovation through AI easier. “By offering an end-to-end AI platform that makes it even easier to ideate, conceptualize, develop, deploy and iterate, Google has moved us even further in democratizing and enabling innovation,” said Lin Classon, head of product at the managed service provider Ensono LP.

Building AI

The AutoML Tables tool is actually just one of several to be included in Google’s new AI Platform, currently available in beta, which is billed as an end-to-end development platform that helps teams prepare, build, run and manage their machine learning projects.

Google offers a wide range of machine learning tools out of the box, but it can’t possibly cater to everyone’s needs, so AI Platform is designed to give companies everything they need to build their own.

“With AI Platform, you can ingest streaming or batch data, and use a built-in labeling service to easily label training data — like images, videos, audio, and text — by applying classification, object detection, entity extraction, and other processes,” Rajen Sheth, director of product management at Google, wrote in a second blog post. “You can import your data directly into AutoML, or use Cloud Machine Learning Engine, now part of AI Platform, to train and serve your own custom-built ML models on GCP.”

Mueller said the AI Platform looks like it’s designed to bring all of Google’s AI products and services under a single umbrella, which is being done to encourage more “regular” developers to become AI developers.

Google is also updating its Cloud AutoML developer toolkit, which provides a drag-and-drop interface for training AI models, and was first introduced last year. As well as the new AutoML Tables tool described above, Cloud AutoML is getting an updated AutoML Vision service for image recognition.

It now can optimize machine learning models to run on so-called “edge” devices such as connected sensors and smart cameras. Adding intelligence to such devices isn’t easy because they’re often plagued with issues such as unreliable connectivity and latency, Sheth said. AutoML Vision Edge helps to remedy that by simplifying training and deployment, he added.

Cloud AutoML gains another new service in AutoML Video, which helps developers to create custom models that automatically classify video content with pre-defined labels. “This means media and entertainment businesses can simplify tasks like automatically removing commercials or creating highlight reels, and other industries can apply it to their own specific video analysis needs — for example, better understanding traffic patterns or overseeing manufacturing processes,” Sheth said.

Lastly, Google is also boosting the infrastructure it provides to run its AI services. The company is making its third generation of liquid-cooled cloud Tensor Processing Units, which are AI accelerator application-specific integrated circuits, generally available.

These, together with Google’s older TPUs, can now be used with the Google Kubernetes Engine service in order to run containerized machine learning workloads. In addition, Google is now offering access to Nvidia Corp.’s new Tesla T4 chips in eight of its cloud regions.

Photo: SiliconANGLE