New Google Cloud AI Platform unifies AI development tools

Google's newly unveiled AI Platform offers AI developers a collaborative environment to test, train and deploy machine learning and deep learning models.

With the Google Cloud AI Platform, the tech giant is attempting to simplify the workloads of AI developers by providing a more unified and collaborative platform for some of its cloud-based AI tools and services.

The Google Cloud AI Platform, now out in beta, enables development, data science and data engineering teams -- working together -- to create and deploy machine learning and deep learning models.

New and old, combined

Unveiled at Google Cloud Next '19 in San Francisco, the platform offers a variety of tools to test and train models. It appears to be comprised of both new and rebranded AI tools and services, according to Bob O'Donnell, president and chief analyst at TECHnalysis Research.

With the unified offering, Google is "further extending [its] AI capabilities and reaching out to companies that don't have the in-house expertise" to begin creating AI models on their own, O'Donnell said. "[AI Platform] gives people a whole bunch of tools and capabilities to get started. It's hard to get started -- that's one of the big challenges."

In addition to enabling teams to create and train models from scratch, the Google Cloud AI Platform offers many adaptable, prebuilt machine learning pipelines and notebooks via the AI Hub, a marketplace for such resources that Google introduced last year.

The platform supports Kubeflow, Google's open source tool for efficiently deploying scalable machine learning workflows on Kubernetes. It also supports Kubernetes-powered Anthos, formerly the Google Cloud Services Platform.

More options, more competitive

[AI Platform] gives people a whole bunch of tools and capabilities to get started. It's hard to get started -- that's one of the big challenges.
Bob O'DonnellPresident and chief analyst, TECHnalysis Research

O'Donnell said this puts Google in a more competitive position against companies with similar products, such as AWS or Microsoft, because customers might not have to fear being locked into a cloud service.

"It's a great way for them to provide an entry point [into enterprise-level AI development]" O'Donnell said. "I think that gives them a great entrée into companies that may not have considered them."

He added that he also thinks there will "be a lot of interest in the ability to migrate across clouds that their Anthos [platform] brings."

Google also introduced several other cloud-based AI products at Google Cloud Next '19, including updates that add new capabilities to its Cloud AutoML and BigQuery ML machine learning platforms.

Among those products was Document Understanding AI, now in beta -- a product that "helps organizations automatically extract insights from documents, but also make your business processes easier," Rajen Sheth, director of product management for Google Cloud AI, said during a livestreamed keynote session from the conference.

Rajen Sheth, Google Cloud Next '19, Google Cloud AI Platform
Rajen Sheth, director of product management for Google Cloud AI, reveals AI Platform at Google Cloud Next '19

Google also introduced Google Cloud for Retail, a unified AI system for retailers that combines Google's Recommendations AI, AutoML Tables and Vision Product Search services.

The company also released updates to its Contact Center AI, unveiled last year, including a new agent assist tool that automatically feeds relevant customer service information and workflows to humans for complex calls and enhanced automated voice profiles.

Dig Deeper on AI infrastructure

Business Analytics
CIO
Data Management
ERP
Close