Focus on enterprise AI applications, Google Cloud Computing Conference releases new products such as Google Vids and Gemini Code Assistant

Wallstreetcn
2024.04.09 17:58
portai
I'm PortAI, I can summarize articles.

Google announced a series of artificial intelligence product updates for enterprise customers at the cloud computing conference, including Gemini Code Assistant, Google Vids, Vertex AI Agent Builder, and other new products. Gemini Code Assistant will compete with GitHub's Copilot Enterprise Edition, Google Vids is an AI-driven video creation tool, and Vertex AI Agent Builder can help enterprises build AI agents. In addition, other new products were also launched. Google emphasized the security and usability of its AI products for enterprises

Google held its annual cloud computing conference in Las Vegas, USA on Tuesday, releasing a series of AI product updates for cloud computing enterprise customers, covering various aspects from Gemini to AI development, operations, and security.

Google emphasized that its AI products are secure and usable for enterprises. Thomas Kurian, CEO of Google Cloud, said, "Enterprises have been experimenting with many scenarios of generative AI with us, and now they are deploying them in production, such as capabilities like grounding and improving answer accuracy. People have gradually adapted, seen value, and therefore deployed them."

Google Vids

Currently, large tech companies are seeking to use AI to help customers develop creative content. On Tuesday, Google launched Google Vids, a new AI-driven video creation tool added to Google Workspace.

Google stated that users can create videos while using other Workspace tools such as documents and spreadsheets. Google Vids can handle editing, writing, and production. Users can also collaborate with colleagues in real-time in Google Vids.

Vertex AI Agent Builder

Google also released a new AI tool, Vertex AI Agent Builder, to help companies build AI agents.

Thomas Kurian, CEO of Google Cloud, said, "Vertex AI Agent Builder allows people to easily and quickly build AI conversational agents. You can build and deploy ready-to-use, generative AI-driven conversational agents and guide them like guiding humans to improve the quality and accuracy of model responses."

For this, Google uses a process called "grounding" to associate the answers made by AI with reliable sources of information. In this case, it relies on Google Search (which may actually be accurate or inaccurate).

Gemini in Databases

Google stated that Gemini in Databases has a range of features that can "simplify all aspects of database operations." In simple terms, this is an AI-driven tool for Google Cloud customers to create, monitor, and migrate application databases, aimed at developers

AI Introduces Network Security Tools

Google has joined the ranks of productizing generative AI into security tools, launching a variety of new products and features for large companies, including training to identify threat information and analyze large amounts of potential malicious code. It also allows users to search for ongoing threats or danger signs using natural language.

Another product is Chronicle, a network security telemetry service provided by Google for cloud customers to assist in network security investigations. The third product is the Security Command Center, an enterprise network security and risk management suite.

Gemini Code Assistant

Google has released its enterprise-focused AI code completion and assistance tool, Gemini Code Assist, which will be provided through plugins for popular editors such as VS Code and JetBrains.

Compared to Google's previous generation programming assistant Duet AI, Code Assist competes more directly with GitHub's Copilot Enterprise rather than the basic version of Copilot. This is thanks to some Google-specific improvements.

This includes support for Gemini 1.5 Pro, which is known for its one million-token context window, allowing Google's tools to introduce more context than competitors. Google states that this means more accurate code suggestions, such as inferring and changing large blocks of code.

Like GitHub Enterprise, Code Assist can also be fine-tuned based on a company's internal codebase.

NVIDIA's Blackwell Platform

NVIDIA's next-generation Blackwell platform is expected to be introduced to Google Cloud in early 2025. While this may sound distant, there are some exciting features to look forward to: support for high-performance NVIDIA HGX B200 for AI and high-performance computing workloads, as well as GB200 NBL72 for training large language models (LLMs). Additionally, the GB200 servers will utilize liquid cooling technology.

Google Workspace

Google has introduced some new features for Workspace, including voice prompts for the AI-based "Smart Compose" feature in Gmail on mobile. Another feature of Gmail includes a way to instantly convert rough email drafts into more refined emails On Sheets, users can send customizable alerts when a certain field changes. Additionally, new templates make it easier to create new spreadsheets. For Doc users, there is now support for tabs, allowing users to "organize information in a single document instead of linking to multiple documents or searching in Drive."

Google also plans to charge for two new AI features in the Google Workspace productivity suite, at $10 per user per month. One will be a new AI meeting and messaging add-on that can take notes for users, generate meeting summaries, and translate content into 69 languages. The other is a newly introduced AI security package to help administrators keep Google Workspace content more secure.

Imagen 2

Google previously released an AI image generator that was taken down shortly after injecting gender and race into prompts about individuals. On Tuesday, Google launched an enhanced version of the image generation tool called Imagen 2, which is located within the Vertex AI developer platform and focuses more on enterprise use.

Imagen 2 is now generally available and brings some interesting new features, including inpainting and outpainting. There is also a feature called "text-to-video" where users can now create short, four-second videos from text prompts, similar to AI-driven clip generation tools like Runway, Pika, and Irreverent Labs.

Chrome Enterprise Premium

Google has expanded its Chrome Enterprise product suite with the introduction of Chrome Enterprise Premium.

Google has long provided an enterprise version of the Chrome browser. Through Chrome Enterprise, IT departments can manage employees' browser settings, installed extensions, and the web applications they use. More importantly, they also gain many new security control features, including data loss prevention, malware protection, phishing prevention, and zero-trust access to SaaS applications.

Chrome Enterprise Premium costs $6 per user per month and primarily expands the security features of existing services, as the browser is now the endpoint for much of the high-value work done within companies

Gemini 1.5 Pro

Google released Gemini 1.5 Pro on Tuesday, the most powerful generative AI model from Google is now available for public preview on Vertex AI.

Gemini 1.5 Pro was launched in February, joining Google's Gemini generative AI model family. Undoubtedly, its most notable feature is the amount of context it can handle: from 128,000 tokens to up to 1,000,000 tokens, where a "token" refers to a subpart of the original data (such as the syllables "fan", "tas", and "tic" for the word "fantastic").

One million tokens is equivalent to approximately 700,000 words or about 30,000 lines of code. This is roughly four times the data input capacity of Anthropic's flagship model Claude 3, and about eight times the maximum context of OpenAI's GPT-4 Turbo.

So, what exactly can a 1 million token context window do? Google promises it can do a lot, such as analyzing code repositories, "reasoning" across long documents, and engaging in extended conversations with chatbots