Getty Images/iStockphoto

Google Cloud Next 24 recap: GenAI tools for enterprises

New products introduced at Google Cloud Next stand to provide enterprises with GenAI tools for analytics, data management, AI model building and more.

The speed with which Google Cloud is moving to cement itself as a leader in the AI space is impressive and customers have taken notice.

It's been just eight months since the company's last event, yet there were hundreds of AI product enhancements announced last week at Google Cloud Next in Las Vegas.

Between Vertex, Google's enterprise AI platform that prioritizes openness, and BigQuery, their converged and unified data platform, it's hard to not recognize the organization's leadership in cloud data and AI.

Though there were many updates announced at the event, the major focus was on Vertex. Google continues to evolve this platform, and its openness is a core value proposition that resonates with customers.

AI model building tools

Google Cloud's Model Garden provides access to over 130 enterprise-ready models, including first-party models with Gemini and Gemma, third-party models, and open source models. In this area, there were several announcements that stood out to me:

  • Gemini 1.5 Pro is in public preview, providing up to a 1 million token context window. It can support 11 hours of audio, one hour of video, 30,000 lines of code and 700,000 words. Just massive scale.
  • Imagen 2.0 is now generally available. The text-to-image model provides live image support, and more importantly, watermarking to help protect customers' intellectual property.
  • Google and Anthropic's partnership with model support for Haiku and Sonnet, with Opus coming soon. Anthropic hasn't been shy in promoting recent benchmarks of these models and the tight integration with Google Cloud is great for both companies.

Vertex AI Model Builder provides organizations with the ability to discover, tune, optimize, augment, monitor and deploy models. This is an area where Google focused on extending its machine learning operations services to give organizations greater confidence as they move from development to deployment. Prompt management is an increasingly pivotal area.

We've heard a lot about prompt engineering over the last year as organizations seek more control over model outputs. Google Cloud rolled out prompt management features including version history and custom note fields for individual prompt tracking.

The company also announced a feature that will be embraced by customers: side-by-side comparisons of model responses. This delivers a form of explainability where customers can evaluate response options side-by-side to compare outputs, highlight why one response outperforms another and provide certainty scores to improve accuracy.

AI Agent Builder

The new Vertex AI Agent Builder had me the most excited because of the focus Google Cloud put on making generative AI (GenAI) enterprise ready.

The AI Agent Builder provides developers the tools they need to rapidly build and deploy custom AI agents across several areas of the business, whether that be for customers or employees, to support creativity or code development, or enhance data analysis and improve security. It meets developers where they are in their journeys, whether that's a no-code approach, low-code approach or code-first approaches for complete control.

Customers can think of this area as an extension of Vertex AI Model Builder capabilities. There was a lot of emphasis on reducing hallucinations, improving accuracy, relevancy and context, and overall, improving the reliability of GenAI by building more trust.

The biggest announcement in this area was grounding with Google Search. This is a big deal because it empowers organizations to use Google Search results to ground responses. Of course, organizations will use approaches such as retrieval-augmented generation to incorporate their own enterprise data, but with this new feature, organizations are empowered to incorporate the latest data found in Google Search.

AI-ready data platform

There were several analytics-centric announcements, all of which highlighted a continued differentiator for them: BigQuery. 

Google Cloud views itself as having the most scalable and most converged data platform. It can combine structured and unstructured data in one system in a unified way. This convergence of operational and analytical data enables it to deliver a trusted and secure data foundation with seamless integration of their AI technology -- bringing AI capabilities to the data and applications.

There were several BigQuery updates that support its position as a unified platform for data to AI, including BigQuery Metastore, a managed, scalable runtime metadata service that provides universal table definitions and enforces fine-grained access control policies for analytics and AI runtimes.

There are also new streaming innovations, with BigQuery continuous queries and Apache Kafka for BigQuery, which enable organizations to continuously process data as it arrives. There are now tuning and grounding multimodal LLMs with BigQuery enterprise data. And of course, the integration of Gemini within BigQuery and Looker, which takes data interaction to the next level.

Google Cloud's AI momentum is impressive; the company reported that 90% of GenAI "unicorns" and around 60% of funded GenAI startups are Google Cloud customers.

The eye-opening part to me was when Google Cloud CEO Thomas Kurian stated that many of their AI customers are new to Google Cloud. In other words, their AI leadership allows them to take customers from their competition.

It really highlights a shift we're seeing where organizations want movement and value from AI and GenAI investments, and Google Cloud is proving to be the partner to give it to them.

Mike Leone is a principal analyst at TechTarget's Enterprise Strategy Group, where he covers data, analytics and AI.

Enterprise Strategy Group is a division of TechTarget. Its analysts have business relationships with technology vendors.

Dig Deeper on AI technologies

Business Analytics
Data Management