Google Cloud

AI is presenting a once-in-a-generation opportunity to transform how you work, how you run your business, and what you build for your customers. But the first wave of AI, while promising, has been stuck in silos, unable to orchestrate complex work across an entire organization.

True transformation requires a comprehensive platform that connects to your context, your workflows, and your people. That’s why today, we are proud to introduce Gemini Enterprise: the new front door for AI in the workplace.

Delivering this level of transformation requires a complete, full-stack approach to innovation, and this is where Google leads. Our advantage starts with reliable, purpose-built AI infrastructure, and is powered by the pioneering research of Google DeepMind, and our versatile Gemini family of models. Sundar talks more about our company-wide approach in his blog.

This complete, AI-optimized stack is why nine of the top 10 AI labs and nearly every AI unicorn already use Google Cloud. It’s why 65% of all our customers are using our AI products including: Banco BV, Behr, Box, DBS Bank, Deloitte, Deutsche Telekom, Fairprice Group, the US Department of Energy, and many more around the world. Today, we’re proud to announce more AI wins with Figma, GAP, Gordon Foods, Klarna, Macquarie Bank,Melexis, Mercedes, Signal Iduna, Valiuz, and Virgin Voyages. And we’re excited to be the official Cloud provider of the LA28 Games, where Google Cloud will bring our AI innovations to the Olympic and Paralympic Games.

AI that transforms how you work

Gemini Enterprise brings the best of Google AI to every employee through an intuitive chat interface that acts as a single front door for AI in the workplace. Behind that simple interface, Gemini Enterprise unifies six core components:

The platform is powered by Google’s most advanced Gemini models, creating the brains of the system, providing world-class intelligence for every task.

Through a no-code workbench, any user — from marketing to finance, and any other team — can analyze information and orchestrate agents to automate processes across the organization.

To deliver value from day one, it includes a taskforce of pre-built Google agents for specialized jobs like deep research and data insights, and you can easily augment this with custom agents your teams build or with solutions from our extensive partner ecosystem.

An agent is only as good as its context, so Gemini Enterprise securely connects to your company’s data wherever it lives — from Google Workspace and Microsoft 365 to business applications like Salesforce and SAP.

This is managed with a central governance framework, so you can visualize, secure, and audit all of your agents from one place.

And it is all built on a principle of openness with an ecosystem of over 100,000 partners. This ensures customer choice and fosters innovation.

By bringing all of these components together through a single interface, Gemini Enterprise transforms how teams work. It moves beyond simple tasks to automate entire workflows and drive smarter business outcomes — all on Google’s secure, enterprise-grade architecture.

Some companies offer AI models and toolkits, but they are handing you the pieces, not the platform. They leave your teams to stitch everything together. But you cannot piece together transformation.

That’s exactly what we built with Gemini Enterprise: a complete, AI-optimized platform — from our purpose-built Tensor Processing Units to our world-class Gemini models, all the way to the platform and agents that transform workflows. This is what it takes to deliver a truly unified AI fabric for your business, and it’s why customers are already putting Gemini Enterprise to work:

• Banco BV’s relationship managers used to spend hours doing their own analytics. Now, with the help of Gemini Enterprise, it’s done for them, leaving managers with more time to convert new business.
• Harvey is the leading domain-specific AI for legal and professional services, trusted by Fortune 500 legal teams. Powered by Gemini, lawyers are more efficient across contract analysis, due diligence, compliance, and litigation, saving hours and hours of time.

This isn’t just about making one task easier. It’s about making entire workflows smarter by searching and finding information from all your enterprise documents, applications, email and chat systems, and automating processes using agents with any of your enterprise applications.

Gemini Enterprise highlights our commitment to an open platform – working seamlessly in Microsoft 365 and Sharepoint environments. And when you use Gemini Enterprise with Google Workspace, you get further benefits. Today, we are announcing the first of many multi-modal agents harnessing the power of Gemini to understand and create text, image, video and speech, built right into the Workspace apps you already use:

• Video: With Google Vids, you can now transform one type of information, like a presentation, into a completely different format — an engaging video, complete with an AI-generated script and voiceover. The momentum for Vids has been incredible, with 2.5 million people using it every month.
• Voice: In Google Meet, we are bringing real-time speech translation to all business customers. This goes beyond just words, capturing your natural tone and expression to make conversations seamless, no matter what language you speak. This builds on the voice intelligence from our ‘take notes for me’ feature, which has seen usage grow more than 13x since the beginning of the year.

An agent is only as good as its context. Gemini Enterprise integrates with your organization’s data — wherever it lives — to build that context, and deliver relevant, accurate, and trustworthy results. Today, as part of Gemini Enterprise, we are announcing:

• A new Data Science Agent, in preview, to automate data wrangling and ingestion. It accelerates detailed data exploration, instantly finding patterns and streamlines complex model development by generating multi-step plans for training and inferencing, eliminating manual, iterative fine-tuning.

Customers like Morrisons, Vodafone, and Walmart are already using this agent to accelerate their data workflows and remove friction from the customer experience.

AI that transforms how you run your business

Customer engagement is one of the most critical use cases for AI adoption and our Customer Engagement Suite – our Conversational AI solution for web, mobile apps, call centers and point of sale – works alongside your customer service reps to answer questions via chat and voice and take actions. The business impact is real, and leading companies are seeing results now:

• Commerzbank was an early adopter of Customer Engagement Suite, using it to build Bene, its own specialized chatbot. They are now leveraging Gemini to further enhance the experience, enabling it to handle over two million chats and successfully resolve 70% of all inquiries.
• Mercari, Japan’s largest online marketplace, is overhauling its contact center with Google AI to foster an AI-driven customer service experience, which is projected to yield a 500% ROI by reducing customer service rep workloads by at least 20%.

Today, we are announcing true, next-generation conversational agents, in preview, that connect directly into Gemini Enterprise. These provide more value to you in the following ways:

1. How you build: We are introducing a new, easy-to-use low-code visual builder. You can build a customer engagement agent once, and configure it for all your channels — telephony, web, mobile, email, and chat. These new agents support over 40 languages.
2. The underlying intelligence: These next-gen agents are powered by our latest Gemini models. This means incredible, natural-sounding voices, with the ability to handle accent transitions and real-world noise from a bad phone connection with industry-leading accuracy and latency.
3. Your time-to-value: The new AI augmentation services and prebuilt specialized agents allow agents builders to build, test, deploy and monitor agents faster than ever. In addition, we also use AI assisted coaching to vastly increase the productivity of your employees. This makes your entire contact center — both human and digital — more efficient and effective.
4. Deep enterprise integration. These agents are designed to connect directly into Gemini Enterprise. This unlocks two key advantages: deeper personalization, using real-time context from all your business systems; and unified governance, allowing you to manage all your agents from the same central platform.

AI that transforms what you build

The ultimate transformation is when you use AI to create entirely new experiences for your customers. This starts with empowering the developers who are building your agents and applications. In just three months since launch, over one million developers are already building with Gemini CLI, an AI agent that lets developers interact with Google’s Gemini models directly from a terminal for task automation, code generation, and research using natural language. It has become an essential tool for developers around the world, whose workflows are becoming more complex every day. The best AI shouldn’t force you to switch contexts, it should adapt to your toolchain.

That’s why we introduced Gemini CLI extensions — a new framework to customize your command-line AI and connect it to the services you rely on most. This allows you to build a more intelligent, personalized workflow with a growing ecosystem of extensions from Google, and industry leaders like Atlassian, GitLab, MongoDB, Postman, Shopify, Stripe, and more. It turns your CLI from a simple tool into a personalized command center.

Innovation with agents is leading to an entirely new agent economy, where developers, ISVs and partners can build and earn revenue from specialized agents that communicate and transact with one another. To enable this, we have worked with the industry on an open standard called the Agent2Agent Protocol (A2A), which along with Model Context Protocol (MCP), sets the standard for how agents communicate.

But for agents to be truly autonomous, they must be able to transact. To provide a secure and auditable way for agents to complete payments, last month we announced a new, open protocol: the Agent Payments Protocol (AP2). This is a first-of-its-kind effort, developed with over 100 payment and technology partners, like American Express, Coinbase, Intuit, Mastercard, PayPal, ServiceNow, and Salesforce to establish how agents securely enable financial transactions.

By working with our partners and the larger community to build standardized protocols for key aspects, such as context, communication and commerce, we are laying the foundation for the agent economy.

Our customers are also building our Gemini models directly into their products:

• Klarna is using tools like Gemini and Veo to create bespoke lookbooks that are dynamic, personalized and impactful with shoppers, increasing orders by 50%.
• Mercedes-Benz builds cars with Google AI that can talk to their drivers. They are using Google Cloud’s Automotive AI Agent, which is built using Gemini on Vertex AI to power their MBUX Virtual Assistant, which enables natural conversations and provides personalized answers to drivers for things like navigation and points of interest.
• Swarovski creates personalized customer experience with Vertex AI resulting in a 17% increase in email open rates and 10x faster campaign localization.

This transformation goes beyond code. Our Gemini family of models have been used to create over 13 billion images and 230 million videos. For example:

• Figma is helping their community create more than ever. With tools across their platform powered by Gemini’s Flash 2.5 Image model (more commonly known as: Nano Banana), their users can now make high quality, brand-approved images with just a prompt, edit details with AI, and get all the variety their project needs.
• Virgin Voyages is using Veo’s “text-to-video” and Imagen to create thousands of hyper-personalized advertisements and emails. Each one perfectly matches Virgin Voyages unique brand voice at a scale that would be impossible just a year ago.

The number of customers using our AI models in Vertex continues to grow, including top brands, like: Adobe, Cathay Pacific, Kraft-Heinz, LATAM Airlines, Toyota, Unilever, and more.

The future of AI must be open

Google Cloud’s AI strategy is built on a foundational belief: The future of AI requires an open, collaborative partner ecosystem to ensure customer choice. To make this real, we have built a comprehensive agentic AI ecosystem with more than 100,000 partners supporting every layer of our AI stack – AI infrastructure, AI tooling, ISVs, and services partners. Today, we are advancing this ecosystem in four critical ways:

1. Expanded cross-platform workflows: We’re expanding our work with partners like Box, OpenText, ServiceNow, and Workday – tools you use everyday – to enable sophisticated, cross-platform workflows right out of the box.
2. Scaling with partners: Industry-leading partners, including BCG, Capgemini, HCLTech, Infosys, McKinsey, TCS, Wipro, and many others can assist with planning, deployment, and custom agent development to speed your adoption of Gemini Enterprise. And Accenture, Cognizant, Deloitte, KPMG, and PwC are making announcements today on their internal adoption and expanded services for Gemini Enterprise.
3. Discover validated agents: Customers can now use a new AI agent finder to discover the right agent for your needs, where you have assurance that the thousands of agents you can now find, filter, and deploy have been reviewed for security and interoperability.
4. Market & monetize agents: For those partners building AI agents, we provide you with a simple, powerful way to market and earn revenue from your solutions, instantly connecting you with millions of Google Cloud customers.

Enabling your transformation

Delivering this level of transformation requires a commitment to upskilling your teams. So today, we are announcing a comprehensive set of programs to help you succeed.

To upskill your entire workforce, we are introducing Google Skills, our new platform where training from across Google – from Gemini Enterprise to Google Deepmind – is available for free. On this platform, we’re announcing the Gemini Enterprise Agent Ready (GEAR) program, a new educational sprint designed to empower one million developers to build and deploy agents. Click here to be the first to use Google Skills and learn more about GEAR.

For organizations that want our experts embedded side-by-side with your teams, we are proud to announce a new team – Delta – an elite group of Google AI engineers to help you tackle your most complex challenges.

These programs are all designed to do one thing: help you and your teams build your future with AI.

Your foundation for the future

As AI transforms organizations around the world, Google is the only partner with the full set of offerings that you can tailor to your organization’s needs. And most importantly, we are delivering real business value to help you drive ROI from your AI investments.

This is the power of Gemini Enterprise: the new front door for AI in the workplace. We’re bringing the best of Google AI to every employee, for every workflow. And we’re excited to support you every step of the way.

From the classroom to the boardroom, the world of work is shifting at an incredible pace. As advancements in AI and cloud computing gather speed, it’s not just about adapting — it’s about discovering powerful new ways to thrive, regardless of your role.

To help everyone keep up, we’re announcing three major updates to our learning programs, timed with the launch today of Gemini Enterprise.

• A new platform, Google Skills, that will bring together nearly 3,000 courses and labs in one place — including content from across Google Cloud, Google DeepMind, Grow with Google and Google for Education.
• A new initiative, the Gemini Enterprise Agent Ready (GEAR) program, aims to empower one million developers to start building enterprise-ready AI agents with our new Gemini Enterprise platform.
• A partnership with the financial services company Jack Henry to help hire skilled individuals quickly as part of our skills-based hiring initiative.

With these updates, we’re delivering on our commitment to help everyone access the AI learning they need — including our Google Cloud customers in search of skilled developers.

What’s new for Google Cloud learners, customers and partners on Google Skills?

To support people at all skill levels — from students to developers to executives — we’re introducing Google Skills. This new learning platform is designed to help people develop the skills they need to be successful in today’s job landscape, and to enable businesses to find and develop the talent they need to thrive. In the last year alone, people have completed more than 26 million courses, labs and credentials — and now they’re all in one place.

In addition to content from across Google — like new AI research courses from Google DeepMind — this launch brings with it a host of new content for our Cloud customers to keep them on the cutting edge of AI skill building:

1. Gemini Code Assist: AI-powered learning and new skill badges

Gemini Code Assist will help engineers, developers, data scientists and more jump right into coding with Gemini without leaving the Google Skills platform. It’s been enabled in more than 20 hands-on labs and will be part of all relevant labs going forward. And developers can also prove Gemini Code Assist skills by earning a new skill badge: Kickstarting Application Development with Gemini Code Assist.

We’re also meeting the vibe-coding moment with two new skill badges for app devs: Building a Smart Cloud Application with Vibe Coding & MCP establishes foundational knowledge and practical skills in Model Context Protocol server development and vibe coding on Google Cloud. Deploy an Agent with Agent Development Kit empowers devs to build advanced AI systems where different AI parts work together smoothly, using common methods like the Model Context Protocol, Agent-to-Agent protocol and Agent Development Kit.

1. A better experience for Google Cloud customers and organizations

You asked. We listened. As part of Google Skills, we’ve kept the best of Google Cloud Skills Boost, while adding features you’ve been waiting for on the all-new Google Skills platform. For example, Cloud customers will continue to have access to the entire Google Cloud content library for free, now including new content from Google DeepMind.

We’ve also added a new feature that lets you assign the most relevant courses and hands-on labs to your teams. This personalization benefits your business and accelerates your team’s ability to innovate.

Plus, you can now use company leaderboards — customized for your organization — that spark friendly competition and add a dose of fun. And finally, for Google Skills admins, we’ll be adding features to make reporting more advanced. That means more data, more flexibility and more insights that let you keep track of your team’s progress in real time.

1. Cloud learning has never been more fun: new gamified features

For organizations, Google Skills makes it easier to keep your teams engaged and up to date. This is key: 74% of decision makers agree that technical learning resources improve employee productivity, and 71% of organizations realize revenue gains after engaging with these resources.¹ In other words, more learning is a universal boon to business.

But learning alone isn’t enough. It needs to be fun, engaging and easy to fit into your schedule. That’s why we’re bringing new AI-powered and gamification features to Google Skills to make learning more effective.

Gamified features:

• Leagues: Encourage friendly competition based on earned points with this dashboard widget.
• Achievements: Celebrate your learning milestones with new visuals and easy options for social sharing.
• Learning streaks: Promote consistent learning habits through bonus points and “streak freezes.”

GEAR: A new sprint to empower one million developers

Today, we’re also announcing the Gemini Enterprise Agent Ready (GEAR) program, a new educational sprint designed to empower one million developers on our new agentic platform, Gemini Enterprise. Through a cohort-based approach, we will help them build, deploy and manage agents, and as part of GEAR, developers can earn skill badges through Google Skills.

A direct path to employment; a faster way to hire

Building skills isn’t just good for individual careers — it’s also good for business. That’s why with Google Skills, we’re making it easier for people to get the skills employers are looking for, and for companies to find the talent they need to succeed.

Today, we’re announcing that for those who complete a Google Cloud Certificate in cybersecurity or data analytics in the U.S., there’s now a direct pathway to an interview with leading financial services firm Jack Henry. You’ll get to complete custom, hands-on labs that simulate the company’s real-world scenarios. These labs act as the first stage of the company’s hiring process, giving you a tangible way to showcase your skills and land a new job. We’re excited to expand the model to more Google Cloud customers in the future.

“We are excited for our collaboration with Google Cloud to reimagine talent acquisition. By leveraging Google Cloud Certificates, we have been able to more effectively identify and recruit top talent, helping to fuel our growth to fill critical skill gaps.” – Holly Novak, Chief People Officer, Jack Henry

This initiative is especially important because businesses are actively looking for talent. A recent study found that 82% prefer to recruit and hire professionals who hold these credentials.²

For Google Cloud customers interested in bringing on skilled individuals to your company, learn more about the program here.

We’re just getting started. Whether you’re an individual looking to get hired or a business leader aiming to upskill your teams, there’s something for everyone to thrive in this new world of AI.

__________________________________________________________________________________________

¹ Google/Ipsos, Cloud Learning Services Market Pulse, Quantitative Portion Fielded Sept-Oct 2024, Qualitative Portion Fielded Oct-Nov 2024, (U.S., U.K., FR, DE, IN, BR, MX, JP, AU/N.Z.)

² Google/Ipsos, Cloud Learning Services Market Pulse, Quantitative Portion Fielded Sept-Oct 2024, Qualitative Portion Fielded Oct-Nov 2024, (U.S., U.K., FR, DE, IN, BR, MX, JP, AU/N.Z.)

Google Cloud believes the future of AI should be open, flexible, and interoperable. Today, with the launch of Gemini Enterprise – our new agentic platform that brings the best of Google AI to every employee, for every workflow – we’re introducing powerful new opportunities for partners to integrate their solutions and bring them to market.

Our AI ecosystem is already thriving, with thousands of partner-built agents available to Google Cloud customers today. More importantly, this curated set of agents has been validated by Google Cloud, ensuring customers have confidence in their quality and security as they use agents to transform their businesses.

Extending partner-built agents in Gemini Enterprise

Our goal is to make Gemini Enterprise the central destination for customers to access the agents they use daily, including those from leading technology and SaaS providers. With the combination of Gemini Enterprise and the Agent2Agent (A2A) protocol, agents can securely communicate and coordinate complex tasks with each other. Some of the partners announcing Gemini Enterprise-compatible agents today include:

• Box: The Box AI agent lets users ask questions, summarize complex documents, extract data from files, and generate new content while respecting existing permissions in Box.
• Dun & Bradstreet: D&B’s Look Up agent uses the D-U-N-S Number, a globally trusted identifier, to unify business data from internal and third-party sources for accurate and efficient integration across enterprise workflows.
• Manhattan Associates: Manhattan’s Solution Navigator agent provides instant answers on Manhattan Active solutions, policies, and operations to accelerate response times and efficiency.
• OpenText: Core Content Aviator simplifies content management, enabling users to search and summarize information with AI assistance, including document generation and multilingual translation.
• Salesforce: Agents built on Agentforce and data from Slack will be accessible to users within Gemini Enterprise, enhancing their AI-powered productivity and business insights.
• S&P Global: S&P’s Data Retrieval agent helps users analyze earnings calls, perform market research, and retrieve financial metrics–all with direct source citations.
• ServiceNow: Enabled by A2A and ServiceNow AI Agent Fabric, ServiceNow AI Agents for Service Observability connects with Google Gemini-powered agents to detect, investigate, and recommend fixes for issues in customer cloud deployments – streamlining incident management, and enabling greater organizational agility.
• Workday: Workday agents, such as its Self-Service Agent, deliver immediate insights and enable quick actions, like flagging potential budget overruns, submitting time off, creating HR cases, managing pay information, and more – all directly within the employee’s flow of work.

These and many other partners have committed to integrate their agents with Gemini Enterprise, including Amplitude, Avalara, CARTO, Cotality, Dynatrace, Elastic, Fullstory, HubSpot, Invideo, Optimizely, Orion by Gravity, Pegasystems, Quantum Metric, Supermetrics, Trase Systems, UiPath, and Vianai.

Discover validated agents using natural language search

In addition to deploying partner agents that integrate with Gemini Enterprise, customers can also now use a new, Gemini-powered AI agent finder – with natural language search – to help discover the right AI agents for their needs. Customers can search for agents from trusted vendors and filter by industry, use case, and whether they have been validated for A2A and deployment to Gemini Enterprise. These agents can then be purchased from Google Cloud Marketplace or directly through partners, and deployed into their environments.

This curated discovery experience is enabled by our enhanced AI agent ecosystem program and a rigorous framework for partners to validate their agents. We’ve also introduced the new “Google Cloud Ready – Gemini Enterprise” designation to recognize agents that meet our highest standards for performance and quality, helping accelerate adoption of trusted solutions and giving partners a new path to monetize their agents.

Growing partner services to help customers succeed with AI

Google Cloud remains partner-led in our professional services approach. With the launch of Gemini Enterprise today, our consulting partners are already expanding their services offerings to help customers accelerate their adoption of AI agents. In fact, many of these partners are already powering their own businesses with Gemini Enterprise. Major expansions include:

• Accenture is driving successful adoption of Google Cloud’s AI technology with clients across industries; expanding agentic capabilities via the Accenture and Google Cloud generative AI Center of Excellence; and launching agents on Google Cloud Marketplace.
• Deloitte is leveraging its “Agent Fleet” of live agents to help clients utilize Gemini Enterprise to deploy industry-tailored agents and co-innovate at scale.
• Capgemini has developed a variety of agents with Google Cloud’s AI technology across sectors, which it will bring to Gemini Enterprise and the Google Cloud Marketplace.
• Cognizant is accelerating agentic AI adoption for its clients and through internal use of Gemini Enterprise and by investing in Google Cloud Centers of Excellence around the globe.
• GlobalLogic, a Hitachi Group Company, will adopt Gemini Enterprise internally and provide digital engineering services to accelerate customer adoption, including building AI Agents securely and at scale.
• KPMG is using Gemini Enterprise to enhance the speed, accuracy, and quality of client delivery and to elevate the employee experience at KPMG with AI and agents, making everyday work easier.
• PwC is advancing client AI transformation by combining its agent OS technology with Gemini Enterprise, including the deployment of agents it has used successfully internally.

Get started bringing agents to Gemini Enterprise

Gemini Enterprise is available today and includes access to thousands of agents from our ecosystem. Partners interested in learning more can visit the AI Agent Program page.

We’re committed to providing partners with the platform and resources needed to scale their AI agents on Gemini Enterprise, and we look forward to seeing the solutions they deliver to customers.

Startups are using agentic AI to automate complex workflows, create novel user experiences, and solve business problems that were once considered technically impossible. Still, charting the optimal path forward — especially with the integration of AI agents — often presents significant technical complexity

To help startups navigate this new landscape, we’re launching our Startup technical guide: AI agents. It provides a systematic, operations-driven roadmap for embracing the potential of agentic systems.

What does this potential look like? AI agents combine the intelligence of advanced AI models with access to tools so they can take actions on your behalf, under your control. Unlike traditional AI, agentic AI can break down intricate tasks, refine plans, and dynamically utilize external resources and tools. The key takeaway is that AI agents can tackle complex, multi-step problems, ultimately transforming from a passive tool into a proactive problem-solver.

If your startup is looking to get in on the agentic AI action, here are some initial steps to consider. And when you’re ready to get building, you can get more details in our guide or even reach out to one of our AI experts at Google Cloud.

Choose your path: Build, use, or integrate

Every startup’s journey is unique, which is why Google Cloud offers a flexible agent ecosystem that supports the comprehensive development of agentic systems. You can:

• Build your own agents: For teams that require a high degree of control over agent behavior, the open-source Agent Development Kit (ADK) is your go-to development framework. ADK is built for a custom, code-first approach, empowering developers to build, manage, evaluate, and deploy AI-powered agents. For an application-first approach, Google Agentspace orchestrates your entire AI workforce and empowers non-technical team members to build custom agents using a no-code designer.
• Use Google Cloud agents: With rapid prototyping and easy ways to integrate AI into your existing apps, managed agents let you focus on core business logic rather than managing infrastructure. Gemini Code Assist is an AI-powered assistant for developers, while Gemini Cloud Assist is an AI expert for your Google Cloud environment.
• Bring in partner agents: For more specialized use cases, you can easily integrate third-party or open-source agents into your stack via the Google Cloud Marketplace. You can also explore the Agent Garden to deploy prebuilt ADK agents that already support data reasoning and inter-agent collaboration.

No matter which path you choose, our ecosystem is designed for interoperability, built on open standards like the Model Context Protocol (MCP) and the Agent2Agent (A2A) protocol.

4 steps for building your first agent

The Startups technical guide: AI agents provides a complete roadmap for building production-ready AI agents. Here’s four core steps we’ve identified that can help define your first agent, using Agent Development Kit (ADK).

Step 1: Give your agent an identity

First, define your agent’s core identity. You’ll want to give it a unique name for logging and delegation, a clear description of its capabilities so other agents can route tasks to it, and identify the right AI foundation model (like Gemini 2.5 Pro or Gemma) to power its reasoning. Precision here is critical. The model you’re using treats every part of this definition as a prompt, and vague descriptions can lead to “context poisoning,” causing the agent to pursue incorrect goals.

Step 2: Write the “prime directive” with instructions

Next, give your agent its “prime directive” using the instruction parameter. This is where you define its persona, core objectives, and do’s and don’ts. Effective instructions clearly specify the desired outcomes for your agent, provide examples for complex tasks (e.g. few-shot prompting), and guide the agent on how to use its tools.

Step 3: Grant superpowers with tools

Transform your agent from a pure conversationalist into a system that can take action by equipping it with functions to call external APIs, search databases, or interact with other systems. In doing so, you grant it broader capabilities. For example, a bug assistant agent uses tools to fetch user details from a CRM or create a ticket in a project management system. Since the agent chooses which tool to use based on its name and description, making them clear and unique is crucial to avoid looping behaviors or incorrect actions.

Step 4: Master the lifecycle: test, deploy, operate

Building an agent is a continuous cycle, not a one-off task. Because agentic systems are non-deterministic, standard unit tests are insufficient. Our guide shows you how to evaluate an agent’s “trajectory” — its step-by-step reasoning — to ensure quality and reliability. This operational rigor, which we call AgentOps, is key to confidently deploying your agent on platforms like Vertex AI Agent Engine or Cloud Run and operating it safely in production.

Agents already in action

Startups are constantly innovating their agentic journeys , here’s a look at two startups that use Google Cloud’s models and architecture to run their agentic systems:

Actionable insights for better employee engagement

Wotter, a provider of next-generation Employee Engagement solutions, seeks to better understand what employees want and empower organizations with the insights they need to get the best out of their people by asking the right question to the right person at the right time.

Gemini 2.5 Flash was the right foundation model for Wotter’s smart assistant, blending speed with long-context reasoning. Wotter’s Flash models use agentic methods to manage extensive and ongoing sources of data, such as employee interactions and feedback, while still responding to queries on this data in seconds – and at a lower cost per query.

Eliminate a long-standing legal industry pain point

As people in the legal industry know too well, complex document reviews can ruin nights and weekends while turning focus away from strategic work. Enter Harvey, which is equipping legal professionals with domain-specific AI to maximize efficiency and keep legal professionals’ attention on activities that move the needle for their firms and clients.

Harvey evaluated several foundation models and ultimately found that Gemini 2.5 Pro achieved the leading score of 85.02% on its BigLaw Bench benchmark, the first of its kind to represent complex legal tasks. Gemini 2.5 Pro showcased strong reasoning across inputs consisting of hundreds of pages of materials—a common scenario in legal work. The model then used these materials to generate longer-form and comprehensive outputs, enabling deeper insights and analyses.

These core capabilities proved Gemini 2.5 Pro’s potential across complex legal work that requires reasoning over large sets of documents to support diligence, review, and use case drafting. Further, Vertex AI provides the stringent security and privacy guarantees that build trust in the Harvey platform among clientele. Gemini and Vertex AI are now an important part of Harvey’s vision for future product development.

Build what’s next, faster

The Startup technical guide: AI agents provides the blueprint your team needs to turn your vision into a production-ready reality. By using a code-first framework like ADK and the operational principles in this guide, you can move beyond informal “vibe-testing” to a rigorous, reliable process for building and managing your agent’s entire lifecycle. For your startup, this disciplined approach becomes a powerful competitive advantage.

No matter where you are with AI adoption, we are here to help. Contact our Startup team today and you could get up to $350,000 USD in cloud credits with the Google for Startups Cloud Program,

We recently hosted our first-ever AI Builders Forum, where we gathered with hundreds of the top founders, VCs, advisors, researchers, and teams powering the startups who are building the future with AI. And now, this week, many of us are together again in San Francisco for Tech Week.

A big reason for us to gather with our peers and the next generation of groundbreaking companies is that at Google Cloud, we’ve always worked to provide companies with the very same tools and infrastructure that power Google’s own services and drive our continued growth. In recent years, we have been able to take that to an entirely new level, making our leading generative AI technology available to even the youngest startups.

During the AI Builders Forum, we showcased the work of dozens of startups who have taken that technology in new and exciting directions. There’s so many more hard at work with Google Cloud that we wanted to highlight as many as we could. Here, we’ve gathered 150 to be exact, breaking them out by the sectors and specialties in which they’re creating exciting new offerings that are changing the way we work, create, collaborate, and cure.

For those looking to join their ranks, we encourage you to join our Google for Startups Cloud program, with access to valuable credits and advice from Google experts and experienced VCs. And when it comes to the especially hot segment of AI agents, we’ve just introduced our Startup Technical Guide: AI Agents. It’s an in-depth collection of best practices and step-by-step guidance for building and scaling agentic AI.

We hope this list serves as inspiration for your own work — and that you’ll find yourself being recognized soon enough, too. All it takes is the right AI, the right partners, a good team, and a great idea.

• Arize AI, an AI agent and engineering platform, partners with Google Cloud to help organizations successfully develop, evaluate, and observe their generative AI applications. The Arize AX platform seamlessly integrates with Vertex AI models and runs on Google Kubernetes Engine (GKE), which allows a very lean operations team to easily scale services and provide deep visibility into every layer of AI systems.
• Augment Code, an AI coding assistant, integrated Anthropic’s Claude 3.5 Sonnet via Vertex AI to power its codebase chat feature. This delivered an instant improvement to chat performance after quick implementation, enhancing security and allowing customers to build and troubleshoot code faster.
• Aviator, an engineering productivity platform, uses Google Kubernetes Engine, Vertex AI, and Gemini to scale its engineering productivity platform to thousands of users while accelerating the development of new generative AI features.
• Cursor is an AI-powered code editor that integrates advanced AI features like autonomous agents and codebase-aware chat. It aims to create a highly effective human-AI programmer by automating tasks, understanding entire codebases, and accelerating development velocity, using models like Gemini and Anthropic’s Claude on Google Cloud’s Vertex AI.

• DeepSource, a platform for code quality and security, leverages Gemini and Google Kubernetes Engine (GKE) to help developers automatically analyze and remediate code. The platform uses Gemini-powered AI agents for its Autofix™ remediation engine, which increases the accuracy of its static analysis and provides automated fixes. Running on GKE, DeepSource can automatically scale to process tens of millions of lines of code per day, reducing operational costs and accelerating time to market for over 6,000 companies.
• Factory AI, a platform for agent-driven software development, accelerates engineering by unifying context from sources like GitHub and Jira to delegate tasks like feature development and migrations. It uses Gemini 2.5 Flash for data ingestion and Gemini 2.5 Pro for advanced code/document generation.
• Fireworks AI, a generative AI platform, uses Google Kubernetes Engine and Compute Engine to run its fast and efficient inference engine. This enables the company to process over 140 billion tokens daily, offering customers high uptime and throughput with lower latency and costs.
• Lovable, an AI software engineer, leverages Google Cloud’s Vertex AI to deploy and unify its core language models, including Gemini and Anthropic’s models. This unique orchestration enables users to create complete, full-stack web applications from plain English descriptions, effectively cutting the time required for prototype development from weeks to minutes and app development from months to hours.
• Naologic, an AI application platform, uses Gemini APIs, Google Kubernetes Engine , and MongoDB Atlas on Google Cloud to build apps on top of legacy ERPs. The solution delivers fast query responses regardless of complexity, enabling powerful, natural-language chat and scaling for complex AI workloads like RAG and image search.
• Qodo addresses critical code quality concerns with an agentic platform that works within existing developer workflows. Qodo integrates into Vertex AI Model Garden and provides automated pull request reviews at no cost to open-source projects.
• Resolve AI, an always-on AI SRE, autonomously investigates incidents and helps run production systems using code, infrastructure, and observability data. With the intelligence and performance of Gemini on Vertex AI, Resolve AI improves MTTR, reliability, and engineering velocity for its customers.
• Sieve, an AI research lab focused on video data, builds multimodal AI systems to automate dataset creation, improve data quality, and provide relevant metadata that powers frontier model training. Sieve uses Vertex AI for large-scale video processing, content moderation, and indexing, including tasks like text recognition, captioning, and enrichment.
• Systalyze, an enterprise AI deployment platform, reduces the cost and complexity of AI. Through a partnership with Google Cloud, it reduces deployment costs by up to 90% and accelerates fine-tuning, inference, and agentic AI by 2–15x, while keeping data fully private and secure.
• Turbopuffer, a startup offering serverless vector and full-text search, uses Google Cloud Storage, Google Kubernetes Engine, and Google Compute Engine to help AI businesses overcome the high costs and complexity of traditional database architectures. ​Its solution has reduced AI database cost by up to 90% for customers, manages over 1 trillion documents, and handles more than 10 million writes and 10,000 queries every second.
• Vercel democratizes access to AI models through its AI SDK and AI Gateway, making it seamless for developers to integrate agentic capabilities into their applications. By providing unified access to leading models like Google’s Gemini through a single interface, Vercel has enabled teams to build AI-powered features faster and more reliably.
• VESSL AI, an MLOps platform, uses Google Cloud to accelerate AI model development and reduce costs for its users. By leveraging Google Kubernetes Engine, VESSL AI can dynamically scale ML workloads, helping users create AI models up to four times faster and realize up to 80% savings on cloud expenditures. The platform’s integration with Vertex AI provides users with access to powerful models and AutoML solutions, further streamlining the MLOps lifecycle.
• Windsurf provides an AI-powered code completion and generation tool for developers. Their AI integrates directly into IDEs, offering suggestions, generating code, and refactoring existing code. Windsurf uses Gemini 2.5 Pro to power its coding assistance IDE, and to support integrations with Cognition’s Devin AI.

• Atropos Health, a healthcare data analytics company, optimized its GENEVA OS to work with Google Cloud’s Healthcare Data Engine (HDE) and BigQuery. This enables customers to efficiently and securely convert data into valuable insights and evidence.
• Cerebra AI develops AI software for analyzing non-contrast CT (NCCT) scans, with a focus on early stroke and cancer detection. It fine-tunes MedGemma on NCCT images and leverages Gemini’s few-shot generalization capabilities to rapidly adapt the model for various diagnostic tasks.
• CitiusTech, a global Healthcare technology services firm, uses Google Cloud to improve patient experience, reduce administrative burden on clinical staff, and save costs for healthcare systems. The company has developed an AI search solution using Vertex AI to efficiently connect patients with the right specialists and automate critical workflows.
• Congruence Therapeutics, a computationally driven biotechnology company, uses its proprietary platform, Revenir, to build a pipeline of small molecule correctors. The platform identifies novel allosteric and cryptic pockets in proteins to rescue aberrant function.
• CoVet is an AI assistant built by veterinary professionals, for veterinary professionals, that uses Gemini, Cloud Functions, and other GCP solutions to help veterinary teams automate administrative work, save hours every day, and refocus on what matters most: exceptional patient care.
• Digital Diagnostics, a healthcare diagnostics company, uses Google Cloud’s secure infrastructure to enhance the reach of LumineticsCore, its AI-powered diagnostic tool for diabetic retinopathy. This approach protects sensitive health data and ensures patient privacy and regulatory compliance.
• DNAstack, a leading genomics data management and analysis platform, leverages Google Cloud’s scalable infrastructure and advanced analytics tools to accelerate research and discovery in personalized medicine.
• Evogene uses Google Cloud and Vertex AI to replace life sciences’ costly “spray and pray” molecular discovery — testing millions of molecules hoping to stumble into effective ones — with their computational platform. They now process 40 billion molecules versus previous millions, while using Vertex AI to develop a cutting-edge small-molecule foundation model that dramatically accelerates drug discovery timelines.
• GenBio AI, a computational biology company, uses Google Cloud to power six specialized AI models in developing AI-driven digital organism simulators. These models simulate biological programming to address critical challenges in medicine and biology.
• Immunai tackles drug development’s decade-long timeline with AMICA, the world’s largest immune-focused single-cell database containing hundreds of millions of cells. Using Google Cloud GPU clusters, they train models that transform complex immune mechanisms into actionable recommendations for 30+ biopharmaceutical partners.
• Infinitus, the first trusted agentic healthcare communications platform, automates clinical and administrative conversations at scale. Our AI agents powered by Gemini’s multimodal capabilities have completed over 5x more conversations than any other solution with payors, patients, and providers to drive revenue and improve health outcomes
• iSono Health, a medical imaging company, developed a Virtual Sonographer, an intelligent, automated 3D Ultrasound platform powered by Google Cloud AI. The platform brings breast imaging directly to the point of care, providing fast, accessible, and repeatable imaging.
• Menten AI, a biotechnology company, uses Google Cloud’s high-performance compute and machine learning capabilities to accelerate the development of peptide therapeutics. This allows the company to rapidly design and optimize novel drug candidates.
• Moonwalk Bio, a preclinical-stage biotechnology company, leverages epigenetic biology and AI to pioneer new medicines for obesity and cardiometabolic disease. Their platform determines the causal relationships between genes and disease pathways for therapeutic targeting.
• Pear Health Labs, a health & fitness AI platform, develops personalized interventions to prevent chronic conditions. It powers recommendations, content search, & dynamic audio coaching. It uses Vertex AI Voice Generation, Vector Search in BigQuery, & the engineering team leverages Gemini Code Assist.
• Sami, a tech-enabled health insurance company in Brazil, built its entire infrastructure on Google Cloud from day one. The company uses AI tools such as Gemini, Notebook LM, and Vertex AI to transform healthcare delivery and accelerate critical processes like prescription validation to seconds. Sami enables information flow across its partners, delivering high-quality, accessible healthcare at scale.
• SandboxAQ is expanding its usage of Google Cloud and running a new AI drug discovery simulation platform on Google Cloud.
• Sully.ai, a healthcare AI company, has built an app store for AI agents designed specifically for healthcare professionals. The platform provides support to clinicians on administrative tasks, so they can focus on patients.
• Tali.ai is the leading medical AI scribe platform, designed to reduce the administrative burden of clinicians. Integrated with multiple EMRs across the U.S. and Canada, it leverages Google’s Vertex and Gemini models to automate clinical note-taking during patient encounters and extract key insights.
• Think Research, a provider of knowledge-based digital health software solutions, uses Google Cloud’s scalable infrastructure and analytics tools to power its platform. This enables the company to deliver more efficient patient care and improve health outcomes.
• Ubie, a healthcare-focused startup founded in Japan, is using Gemini models via Google Cloud to power its physician assistance tool.
• Ufonia helps physicians deliver care Ufonia, a clinical AI company, helps physicians deliver care by using AI to automate clinical consultations with patients. Google Cloud’s full AI stack, including infrastructure, models on Vertex AI Model Garden, BigQuery, and Google Kubernetes Engine, powers its platform by using AI to automate clinical consultations with patients. It is using Google Cloud’s full AI stack to power its platform, including infrastructure, models on Vertex AI Model Garden, BigQuery, and GKE.
• Via Scientific, a bioinformatics company, partners with Google Cloud to deliver Via Foundry, an enterprise-grade platform that uses Gemini and Vertex AI to make the drug discovery process more efficient. The platform transforms complex biological data into actionable insights that can accelerate discoveries.
• Virgo Surgical, a medical video solutions provider, uses Google Cloud Storage and Google Kubernetes Engine to host and process over 1.75 petabytes of video data. This data has been used to create EndoDINO, an AI foundation model for endoscopy that achieves high performance in medical imaging applications.

• Gobii provides AI Agents that automate complex web tasks like forms and workflows directly in the browser. To power these intelligent agents, Gobii leverages Google Cloud, utilizing Vertex AI and our scalable GKE infrastructure.
• Hebbia, a legal and financial research company, integrated Gemini models into its Matrix platform to help organizations build AI agents capable of working across all of their data.
• LiveX, customer service AI agents, uses Google Kubernetes Engine Autopilot and NVIDIA GPUs to power its platform. These technologies reduced total cost of ownership by 50% and operational costs by 66% while supporting an 85% reduction in customer support costs for one of its clients.
• Parallel is using Gemini models to power new products including an API for AI agents to perform high-value tasks using web data.
• Qualia Clear is an agentic system that transforms real estate closings by automating manual title & escrow workflows. It uses tool calling, Gemini 2.5 Flash & Pro, and Google Agent Development Kit to process emails and documents and simplify reporting, improving efficiency and customer service.
• Replicant automates customer conversations for enterprise brands using voice and chat AI agents that replicate the expertise of your very best agents. Gemini helps Replicant deliver consistent service 24/7 that deploys quickly, scales effortlessly and continuously improves to boost ROI and CX.
• Skyvern helps companies automate browser-based workflows with AI. Skyvern uses Large Language Models (LLMs) like Gemini 2.5 Pro and computer vision, to interact with websites, enabling it to automate tasks like filling out forms, procuring materials, and downloading invoices. Skyvern’s AI agents can adapt to website changes, making automation more robust.
• Torq uses agentic AI to automate the entire security operations lifecycle through Socrates, an AI SOC analyst that coordinates specialized agents. Running on Google Cloud’s infrastructure, teams achieve 90% automation of tier-1 analyst tasks auto-remediated without human involvement, 95% decrease in manual tasks and10x faster response times.

• Altumatim, a legal tech startup, uses a platform powered by Gemini on Vertex AI to analyze millions of documents for eDiscovery. This accelerates the process from months to hours, improves accuracy to over 90%, and enables attorneys to focus on building compelling legal arguments.
• Anara, a generative AI research assistant, helps users find and understand scientific documents with verifiable AI summaries and insights. It uses Google Cloud’s scalable infrastructure, AI Studio, and Cloud Functions to power its models and data processing for a global user base.
• Clavata.ai delivers an integrated AI governance and safety platform with intelligent, multi-modal, real-time evaluation engine powered by Gemini. Our tools enable proactive policy enforcement, dynamic debugging, iteration, observability, and problem prevention.
• Harvey, a legal AI company, uses Gemini 2.5 Pro on Vertex AI to automate complex document reviews, a major pain point in the legal industry. The platform provides domain-specific AI that can reason over hundreds of pages of materials, enabling legal professionals to maximize efficiency and focus on strategic work.
• Inspira, a legal tech company, tackles the time-intensive challenge of legal document analysis by providing lawyers with an AI-powered solution built on Google Cloud. Leveraging Gemini, Vertex AI, and BigQuery, Inspira’s platform automates legal document search, analysis, and drafting to reduce workflow times by 80%, allowing lawyers to find answers and relevant decisions in minutes or hours instead of weeks.
• Instalily uses Google Cloud with Gemini 2.5 and Vertex AI to power InstaWorkers™ that transform sales, service, and operations. At a leading field service provider, InstaWorkers™ cut the technician’s diagnosis time from 15 minutes to under 10 seconds, lowered serving costs by 98 percent, and drove a 99 percent improvement in end-to-end workflow speed.
• Markups.ai, an AI contract negotiation agent, turns a days-long human legal review into a minute(s) automated process. By simply emailing a contract, clients receive customized revisions and analysis almost instantly. Gemini 2.5 Pro enabled us to go from handling only first revisions of NDAs, to any revision of any contract (MSAs, DPAs, etc.).
• monday.com, a work management platform trusted by more than 245,000 customers worldwide, leverages Veo to produce training videos, social content, and internal communications in a fraction of the time — empowering all employees, not just designers, to move faster and focus on impact.
• NoBroker, a real estate platform, uses its ConvoZen AI, powered by Gemini and L4 GPUs, to automate customer support across multiple Indian languages. The platform processes 10,000 hours of recordings daily, with AI agents projected to handle 25-40% of future calls and save customers $1 billion annually.
• ObraJobs, a job platform, connects candidates with relevant opportunities, streamline the hiring process, and deliver personalized recommendations for both job seekers and employers. Obra uses Vertex AI to power candidate matching and personalized job recommendations, supported by Cloud Storage, Cloud Run, Cloud Scheduler, Cloud Tasks, and Firestore.
• Provenbase has built its talent recruitment tool for businesses on Google Cloud and is now powering its transformative Deep Search for talent feature using Google Cloud AI.
• Story, an intellectual property startup that powers licensing and monetization services, is working with Google Cloud’s web3 services and infrastructure to bring new capabilities to developers on its platform.
• Upwork, the world’s human and AI-powered work marketplace, connects businesses with independent professionals. By leveraging GCP’s Vertex AI Text to Speech API, Upwork delivers faster, more accurate talent matching and hiring efficiency for clients and freelancers.
• Wotter, an employee engagement platform, uses a Gemini-powered smart assistant and Google Cloud’s robust AI capability to provide real-time insights into employee sentiment. It accurately predicts flight risks and offers actionable “Wott-if” scenarios, enabling leaders to build a data-driven people strategy.

• Bynry‘s SMART360 leverages Google Cloud’s robust infrastructure to empower small and mid-sized utilities to enhance operational efficiency and customer satisfaction.
• Kraftblock, a green tech company, uses Google Cloud Compute Engine to run simulations for its high-temperature thermal energy storage systems, helping energy-intensive industries like steel and ceramics decarbonize. This support helps the green tech startup optimize its solution and scale faster.
• Labellerr, a data labeling engine, uses Vertex AI and Cloud Run to automate annotation and smart QA to help ML teams process millions of images and thousands of hours of videos in just a few weeks.
• Physical Intelligence recently partnered with Google Cloud to support model development, using our secure and scalable AI infrastructure.
• tulanā, an intelligent decision support provider, has a highly customizable platform that uses forecasting, optimization, simulation, and AI to help enterprise clients make better decisions across supply chains and physical infrastructure. tulanā is using Cloud Run to horizontally scale its optimization workloads, Gemini for intelligent ETL processes, and Cloud SQL and Big Query to store customer data.

• Albo, a Mexican neobank, uses Gemini models to power its “Albot” AI chatbot that provides 24/7 financial advice, customer onboarding, and support to millions of first-time banking users. The platform advances financial inclusion while streamlining regulatory compliance and improving operational efficiency.
• Bud Financial, a data intelligence provider focused on banking clients, leverages DataStax Astra DB on Google Cloud with Gemini. The platform enables ultra-fast processing of complex financial data, helping clients reduce fraud by over 90% and shortening the time required to access critical data analytics from weeks to minutes.
• Causal, a financial planning platform, uses Cloud SQL and Gemini models to power its data foundation and accelerate innovation. By offloading database management, the company built an AI-powered wizard that helps users connect data, analyze patterns, and generate financial models in just five minutes.
• CERC, a financial infrastructure company, manages more than 500 million daily transactions using Databricks, BigQuery, and Gemini. This increased processing capacity by 10x without adding to the workforce, allowing the company to process millions in revenue forecasts in just two minutes and accelerate analytics for customers.
• DataCurve, a frontier coding data analytics provider, addresses complex data challenges by combining Web3 and generative AI on Google Cloud. Its platform uses AI agents for deep data analysis and Web3 for data authenticity, delivering insights that help customers take action and improve engagement.
• eToro, a global trading and investing platform, has pioneered a groundbreaking approach to marketing by using Veo to create advertising campaigns. This enables eToro to rapidly generate professional-quality, culturally specific video content across the global markets it serves, which would traditionally require extensive production timelines and significant budgets.
• Fiscal.ai is reinventing financial data infrastructure. Its AI-native platform transforms unstructured public filings into clean, standardized data in minutes, replacing the slow, error-prone legacy of manual aggregation and delivering the mission-critical insights that today’s top investors demand.
• Rogo, a generative AI platform built for the financial industry, uses Google Cloud solutions like Dataflow, Spanner, and Vertex AI to automate complex research and analysis for the world’s leading investment banks and private equity firms. Gemini 2.5 Flash enabled the platform to cut AI modeling time from months to hours and reduce hallucination rates from 34.1% to 3.9%.
• Stax AI, a retirement administration platform, uses Google Cloud’s generative AI and MongoDB to automate data extraction from complex financial documents. The solution processes thousands of brokerage statements in minutes, not hours, helping administrators respond to client inquiries and meet compliance deadlines in a fraction of the time.
• Stream, offers financial tools to employers and employees and is using Gemini models to handle more than 80% of its internal customer inquiries, including questions about pay dates, balances, and more.
• WealthAPI, a German fintech company, uses DataStax Astra DB on Google Cloud with Gemini to deliver real-time financial insights to millions of users. The platform’s scalability allows it to analyze hundreds of thousands of transactions per second and has reduced response times by 90% for its customers.

• Afooga, an AI-powered content experimentation factory, enables businesses to generate, test, and distribute content at massive scale from a single hypothesis, automatically optimizing across TikTok, Meta, YouTube, and more. Afooga leverages Vertex AI and Veo for generative video capabilities, and is architected entirely on Google Cloud.
• Alson AI, a creative platform provider, uses Veo and Gemini to power its creativity platform that helps creators turn ideas into illustrated books and animations, reducing production time from months to minutes and costs from thousands of dollars to $25.
• Cartwheel, a generative animation platform, helps users tell stories faster and more creatively. Its tool uses Gemini Flash for character creation prompts, Imagen for reference-image creation for 3D character development, and Veo 3 for video-to-animation input control that makes output editable by artists.
• ComfyUI, an open-source engine for visual AI, helps creators prototype and automate media generation with pre-set models and more than 20,000 extensions. The platform integrates Gemini 2.5 and Veo 3 for multimodal creation and runs on Google Cloud infrastructure.
• Connected-Stories, an AI creative platform, uses Gemini and Imagen on Vertex AI to help brands overcome manual content personalization challenges. The platform’s AI creative assistant analyzes briefs and turns them into sophisticated strategies, enabling brands to create personalized content at scale and optimize campaigns in real-time.
• fal, a generative media platform for developers, accelerates generative AI model inference to improve the speed with which content is generated. The fal team is working with Google Cloud to leverage its Veo 2 technology to help its users create videos with realistic motion and high-quality output.
• HeyGen, an AI-powered video generation platform, makes creating, translating, and personalizing high-quality videos simple and accessible. HeyGen’s core product leverages Gemini 2.5 Pro, Flash, and Flash-Lite to streamline content creation. With one prompt, HeyGen automates video planning, intelligently analyzes user-generated footage, and optimizes content through advanced visual and audio processing.
• Higgsfield.ai, a generative AI video startup, uses its foundational model on Google Cloud and Vertex AI to enable video creation, helping smaller companies create realistic videos faster and more cost-effectively compared to traditional methods.
• Krea.ai, a creative suite of AI tools, offers real-time image/video generation and personalized model training for artists and marketers. It integrates with Google Cloud, including models as Veo3 and Nano Banana, to provide access to advanced models, enabling users to create high-quality ads, product photos, and game assets.
• Mosaic lets you build and run multimodal video-editing AI agents. A canvas of creative tools becomes your building blocks, enabling simultaneous edits and many versions on autopilot—powered by Gemini 2.5 Pro’s video understanding and Google Cloud (Storage + Cloud Run) for scalable pipelines.
• Nim.video is an AI-first platform for instant short-form video generation from a single prompt. The multimodal platform uses top generative models, including Veo3 and Veo3 Fast, for text-to-video synthesis. It runs on Vertex AI, enabling scaled experiments and orchestration of services like speech recognition and TTS.
• OpenArt empowers social media creators and SMBs to turn ideas into stunning videos in minutes – complete with motion, music, and a narrative arc in one click. Powered by Gemini image models and Google’s Veo3 video model, it makes creating viral posts and brand content fast and effortless.
• Photoroom, provides generative AI photo-editing and design capabilities to consumers and businesses, uses Veo 2 and Imagen 3 to improve the quality of its offering and accelerate its development.
• Prodia offers APIs to integrate generative AI into creative tools. Built on Google Cloud, Prodia relies on GPUs & DWS to serve the fastest text-to-image and instruct-to-edit models in the world, as verified by Artificial Analysis benchmarks. Prodia uses Veo 3 and Nano Banana to further power multimodal AI features.
• Reclip, a “real” social media application, leverages GCP’s Generative Media tools (Veo, Imagen) to create short, engaging animated videos from real time audio clips, captured by their proprietary app. Consumers love sharing these precious, funny and real “Reclips” with their friends and family.
• Scope3 is enhancing its ad-tech platform with AI-powered features, using Gemini 2.5 Flash to offer features like real-time content classification, ensuring content is aligned with brand preference.
• Spot AI, a video AI agent builder, leverages Google Cloud PubSub to turn security cameras into intelligent agents to enhance safety, security, and operations without needing AI/ML teams.
• Wondercraft, an AI-powered content studio that helps users create engaging audio ads, podcasts and more, is leveraging Gemini models for some of its core functionalities and will soon release a Veo 2 integration.
• Writer, an enterprise generative AI platform, builds and trains its 17-plus LLMs on Google Cloud using Google Kubernetes Engine and high-performance NVIDIA GPUs. This allows the company to scale efficiently and cost-effectively, delivering fast, low-latency answers for enterprise customers who demand accurate and on-brand AI-generated content.

• AndesML, a retail media platform, helps large enterprises launch and monetize their own ad networks by showing the right ad to the right customer at the right time. Built on Vertex AI, BigQuery, and Gemini models, the AndesML platform has delivered a 30% performance lift in customer campaigns, accelerated production time by more than 30 days, and reduced operational costs.
• Audiomob, an in-game audio ad platform, replaced its legacy business intelligence system with BigQuery and Looker to gain real-time insights from its global data. This move enabled the company to handle billions of monthly transactions, contributing to triple-digit yearly revenue growth and significant savings in engineering time.
• Hedra, an end-to-end marketing creation platform, is designed to generate high-quality content at scale. Hedra Studio combines its proprietary multimodal models with other leading models like Veo and Imagen, enabling users to produce polished marketing content for any use case. Hedra’s Live Avatars use Gemini to deliver dynamic, real-time interactive video experiences.
• Inworld, an AI platform for builders of consumer applications, uses Google Cloud and Gemini to cost-effectively handle tens of millions of concurrent users with response times measured in milliseconds, meeting strict requirements for quality, cost control, and security.
• Koolio.ai helps creators produce high-quality podcasts and audio content. Koolio.ai integrates Gemini, Lyria, and Veo to power features such as AI-generated dialogue, accurate transcription, intelligent sound effects and music selection, and audio enhancement, streamlining the entire audio creation workflow from concept to final production.
• MNTN uses Google Cloud to power its Connected TV ad platform, making TV campaigns as measurable as search or social. With AI-driven tools like MNTN Matched and Security Command Center, MNTN scales creative and targeting securely and at speed.
• Napster, a metaverse company, is building a no-code 3D e-commerce platform on Google Cloud using Vertex AI and Gemini. This supported 20-85% infrastructure cost reductions and saved over 3,600 developer hours, making immersive 3D web experiences accessible and affordable for its customers.
• Potrero Labs, a creator-focused platform, has launched Jams, an AI-first video social network empowering authentic self-expression. Its platform simplifies video creation, allowing users to record short videos and let Jams enhance them. Jams offers a simple UI with a variety of models under the hood, including Gemini 2.5 Pro for script creation, multi-modal Gemini for video analysis, and Veo 3 for backgrounds, b-rolls, and audio.
• Producer.ai (formerly Riffusion), an AI music platform, trains generative music models and builds products that empower anyone to create the music they imagine. “The Producer” music collaboration agent helps users create original, studio-quality songs from text, audio, or visual prompts. Gemini on Vertex AI assists with prompt augmentation and data pipelines, while Vertex AI APIs offer access to advanced multimodal models for experimentation.
• Rembrand is an AI-powered advertising platform that facilitates in-video product placements for content creators and advertisers on social media and CTV. Powered by Google Cloud’s AI Infrastructure, Rembrand enables brands to genuinely connect with audiences without disrupting the content.
• Satisfi Labs will begin using Gemini models to power a new agentic platform for hundreds of customers in sports, entertainment, and tourism. The Agentic Platform delivers specialized agents for guest experiences, ticketing, on-site, safety, and merchandise tuned by industry experts to help live experience businesses sell more, service faster, and gain real-time insights from every guest conversation
• Scorpion, a digital marketing company for SMBs, uses Google’s VEO AI to scale video ad production. By integrating this technology into its toolkit, Scorpion makes creating professional videos for websites and advertising faster and more accessible for all businesses.
• StatusPro builds NFL virtual reality experiences for gaming and training and is now developing its newest game on Google Cloud, including a new Gemini-powered in-game coach.
• Synthesia, an AI video enterprise platform, helps businesses create instructional videos for employee training, customer support, sales enablement, and product marketing. The company is using Veo 3 to contextually adapt visuals to the content delivered by its AI avatars and voices.
• Tinuiti, a performance marketing agency, used Google’s VertexAI to develop an AI-powered service that develops and optimizes ad copy to increase performance. The tool embodies a philosophy of maximizing growth by minimizing waste, and a recent experiment showed significant ROAS performance improvements compared to human-curated copy.
• Toonsutra, an India-based webcomic platform, is using Google’s Gemini AI to go global. By making stories accessible in regional languages and adding Lyria 2 for music, Gemini for voices, and Veo 3 for animation, they’re creating next-gen immersive comics.
• Velin.ai, a content creation platform for small businesses, offers an AI agent that explains the content and its underlying strategic implications while acting as a unified content workspace. Gemini 2.5 drafts everything from scripts to social campaigns, while Imagen 4 and Veo 3 generate aligned visuals and video clips, ensuring a consistent brand narrative across all content.
• Visla is an AI-powered video creation platform that helps businesses and creators produce pro videos in minutes. Using Google’s Imagen 4, Gemini Flash Image 2.5, Veo 3, and Visla’s AI Video Agent with Avatars, it adapts visuals, narration, and automates polished content for learning, training, and marketing.

• LitLab.ai, a reading platform, leverages Veo3 and Vertex AI to generate personalized, curriculum-aligned stories and provide real-time oral reading analysis. They create decodable content and employ voice recognition for instant teacher feedback on student fluency.
• Savvy revolutionizes learning by automatically generating flashcards and quizzes from PDFs, notes, videos, and podcasts. As students answer, Savvy leverages Gemini to dynamically grade their answers, providing instant feedback and personalized learning.
• Studyhall AI, an AI research platform, graduated from Google Cloud’s UK Growth Accelerator program and built a mobile application that uses Gemini models to help coach students on reading, writing, and exam prep.
• Subject.com, an AI-powered platform for grades 6-12, blends cinematic storytelling with superintelligent AI so students and teachers never get stuck. VertexAI, CloudSQL & BigQuery power Subject’s teacher assistant tool Spark, instant feedback, “ExplainThis” text simplifier, 24/7 Homework Helper, and personalized learning tied to student interests.

• Deeli AI, an AI-powered platform, helps companies discover and evaluate emerging technologies to make informed investment decisions. The company builds its product and data pipeline on various services such as GCE, Cloud Run, and Dataflow, and uses models from the Vertex AI Model Garden.
• Nectar’s AI-driven community agents with Gemini to handle customer conversations on social platforms, influencer marketing, and real-time product feedback at scale for leading brands & retailers. By transforming unstructured social data into actionable insights and powering customer conversations, Nectar helps brands deepen relationships and drive measurable growth.
• Simbe, a multimodal, retail-focused computer vision company, built its AI-powered Store Intelligence platform on Google Cloud. By deploying autonomous robots and sensors, Simbe provides real-time insights into shelf inventory and price accuracy. This helps retailers reduce out-of-stock instances, improve pricing and promotion execution to the high 90% range, and achieve a 4x return on investment within 90 days.
• Zapia AI, a retail technology company, uses AI agents to support millions of users with product discovery, local business searches, and purchase assistance, resulting in over 90% positive user feedback. Its multi-agent orchestration is powered by Gemini to improve agent reasoning, reduce latency, and lower operational costs.
• Zazzle is a global platform for custom products and designs made on demand. Zazzle uses Gemini ADK and CCaaS to facilitate chat-based product discovery and enhance customer experience, making it easier to find the right designs across a wide range of products.

• Aptori, an AI security company, detects vulnerabilities in AI-generated code, prioritizes risks, and automates code fixes in real-time. Aptori uses Gemini to analyze code for security weaknesses and generate context-aware fixes, integrating its AI Security Engineer directly into developer workflows.
• Chainguard, a software supply chain security company, uses Google Cloud Run and Google Kubernetes Engine to provide developers with secure open-source building blocks. Its serverless architecture streamlines operations and product development, reducing infrastructure management costs and enabling them to scale effortlessly to meet increasing user demand.
• Eon, a cloud data protection platform, uses Google Cloud Storage and BigQuery to transform backups into AI-ready data lakes. This allows its customers to eliminate fragmented data silos, reduce secondary storage costs by up to 98%, and improve data recovery times by up to 90%.
• Galileo, an AI observability and evaluation platform for building trustworthy AI applications, addresses the critical challenge of mitigating LLM unpredictability and hallucinations. Using Gemini to build its “evaluation agents” and running on a scalable Google Cloud infrastructure with NVIDIA GPUs, Galileo provides a holistic “trust layer” for reliable AI. This has enabled customers to de-risk over 1,000 AI applications, while go-to-market support from the Google for Startups Cloud Program helps Galileo accelerate growth and unlock new opportunities.
• Prediction Guard is using Google Cloud services like Confidential Computing and Vertex AI to support its platform for added gen AI safety.
• Resistant AI, an AI-powered security company, uses Google Cloud to build solutions that combat fraud in financial services documentation and workflows. Running on Google Cloud infrastructure, the company’s specialized document fraud detectors scrutinize financial documents in 500 different ways, helping to protect automated workflows like those using Google’s Document AI from sophisticated financial crime.
• Specular, an offensive cybersecurity platform, builds AI agents using Gemini 2.5 Pro to automate attack surface management and penetration testing. Their platform automates traditional workflows to identify, assess, and remediate cybersecurity, helping enterprises proactively prioritize and respond to threats.
• Wyze Labs is rolling out new AI-powered anomaly detection features for its security camera systems, powered by Google’s vision AI tool.
• Zefr, a global leader in responsible AI, powers Fortune 500 brand advertising with safety and suitability on platforms like YouTube and TikTok. Using patented Cognitive AI with Gemini Flash and Vertex AI, Zefr analyzes video, image, audio, and text to deliver trusted, scalable solutions.

• aSim, an AI-powered mobile app development tool, allows you to quickly generate, share, and discover mini-apps. Users can instantly generate an app from a prompt, leveraging APIs/LLMs like Google Maps and Gemini, as well as image and video generation from Nano Banana and Veo 3.
• Bud Financial, a data intelligence provider focused on banking clients, leverages DataStax Astra DB on Google Cloud with Gemini. The platform enables ultra-fast processing of complex financial data, helping clients reduce fraud by over 90% and shortening the time required to access critical data analytics from weeks to minutes.
• Bynry’s SMART360 leverages Google Cloud’s robust infrastructure to empower small and mid-sized utilities to enhance operational efficiency and customer satisfaction.
• Citylitics, a predictive intelligence platform, transforms public infrastructure investment for municipalities, utilities, and engineering firms. By automating data processing with Dataflow and Cloud Run, it cuts analysis time by 71% and boosts data sources by 400%, helping customers proactively identify and win new business.
• ContextQA, an Agentic AI software testing platform, enables accurate, user-centric test automation from development start. The platform uses Gemini models to compare application behavior with expected outcomes, adapting automatically to changes.
• Flockx, an AI-powered event app, migrated to Google Cloud to support its growth and control costs. Using Google Kubernetes Engine and autoscaling on Compute Engine, the company accelerated product development and successfully demonstrated scalability by processing over 100,000 AI runs in a single weekend.
• Macro, an AI productivity platform, uses Gemini to modernize knowledge work by offering a unified workspace with features like multi-document chat and editable mind maps. The platform simplifies complex workflows and scales with demand for over 125,000 users in legal, finance, and education, while offering enterprise-grade security, data privacy, and compliance.
• MaestroQA, a conversation analytics data platform, is leveraging Gemini to enhance its AI-powered conversation analytics. By using Gemini, MaestroQA is improving its ability to analyze customer interactions across every channel, providing deeper insights that help businesses boost customer satisfaction and drive growth and retention.
• MLtwist, an AI data pipeline services company, processes, transcribes, translates, and labels large, complex data streams for enterprise applications. It uses Gemini and AI Studio for transcription and labeling tasks, saving approximately 63% of the time required to process even highly illegible documents.
• Moii.AI, a vision AI startup, uses Gemini and BigQuery to analyze CCTV footage for safety and productivity insights. This approach reduced video review time from days to minutes, saved the company $10,000 monthly on AI processing costs, and doubled its engineering team’s efficiency.
• Owl.AI, a sports technology company, delivers AI-powered solutions to professional sports leagues. Their offerings, which include judging and scoring, aim to enhance accuracy, consistency, and eliminate bias. Owl.AI achieves this by leveraging AI models built on Gemini and fine-tuned on Google Cloud to analyze real-time video footage of athletic performances.
• Provenbase has built its talent recruitment tool for businesses on Google Cloud and is now powering its transformative Deep Search for talent feature using Google Cloud AI.
• SandboxAQ is expanding its usage of Google Cloud and running a new AI drug discovery simulation platform on Google Cloud.
• Satlyt, a space compute leader, enables in-orbit AI workloads by orchestrating intersatellite communication and routing. It uses Google Cloud’s Kubernetes Engine, Vertex AI, and scalable data infrastructure to deploy AI agents and plans to deploy Google’s Gemma models in orbit.
• SE3 Labs, a 3D computer vision and AI company, uses Cloud Run to deploy advanced AI model technologies that create “Spatial GPTs,” which are essentially AI models that can understand and interact with the world in 3D.
• Temporal, a durable execution platform for developers, uses Vertex AI to enhance its customer support operations. The solution provides improved visibility into support trends by automatically categorizing 80% of support tickets, allowing the team to anticipate customer needs and identify new opportunities.

As the saying goes, “a penny saved is a penny earned,” and this couldn’t be more true when it comes to cloud infrastructure. In today’s competitive business landscape, you need to maintain the performance to meet your business needs. Luckily, Google Cloud’s Compute Engine and block storage services offer numerous opportunities to reduce costs without sacrificing performance, especially in the context of your migration and modernization initiatives.

In this article, we’ll explore 11 key ways to optimize your infrastructure spending on Google Cloud, from simple adjustments to strategic decisions that can result in significant long-term savings.

1. Choose the right VM instances

One of the most effective ways to reduce Compute Engine costs is to ensure that you’ve properly selected and right-sized your virtual machines (VMs) for their workloads to support your migration and modernization efforts. Whether you’re new to Google Cloud or already using Compute Engine, adopting the latest-generation VMs — such as N4, C4, C4D, and C4A — can deliver substantial savings and improved price-performance.

Powered by Google Cloud’s Titanium architecture, our latest-generation VMs offer faster CPUs, higher memory bandwidth, and more efficient virtualization than their predecessors, so you can handle the same workloads with fewer resources. For existing customers, migrating from older VM generations to the newest VMs can significantly lower total costs while helping you exceed current performance levels. Organizations that have made the switch often report 20–40% better performance along with meaningful reductions in cloud compute spend. For example, Elastic leveraged the general-purpose C4A machine series based on Google Cloud’s Arm-based Axion CPUs, to achieve a compelling efficiency and performance uplift for their workloads.

Beyond general-purpose VMs, we also offer specialized machine types to address unique customer requirements. Compute-optimized HPC VMs like H4D are designed for high-performance computing and data analytics, offering extreme performance for demanding workloads. M4 and X4 instances cater to memory-intensive applications, while Z3 instances are ideal for storage-intensive workloads. Furthermore, if you need complete control over your hardware environment and maximum performance isolation, we offer bare metal instances.

These options help ensure that even the most specialized and performance-sensitive workloads can find an optimal and cost-effective home within the Compute Engine portfolio.

2. Optimize your block storage selections

The best way to lower your block storage TCO, while ensuring your workloads remain successful, is to drive high resource efficiency. Hyperdisk makes it simple to drive high performance and high efficiency by enabling you to optimize your block storage to your workload and through Storage Pools. We’ll discuss each of these capabilities, and how you can use them to lower your block storage TCO below.

Workload Optimized: With Hyperdisk, you can independently tune capacity and performance to match your block storage resources to your workload. Hyperdisk enables you to independently provision performance and capacity at the volume level. You can leverage this capability to purchase just the capacity and performance you need, no more and no less. You can also take advantage of Hyperdisk Balanced’s “baseline” performance (i.e. included free with every volume), you can serve the vast majority of your VMs without purchasing any extra performance.

Storage Pools: Hyperdisk is the only hyperscale cloud block storage to offer thin-provisioned performance and capacity. With Hyperdisk Storage Pools, you can provision the aggregate performance and capacity your workload requires, while still provisioning the volume level capacity performance your workloads request (also known as thin-provisioning). This allows you to pay for the resources you need, not the sum of the volumes you’ve provisioned. As a result, you can lower your overall block storage TCO by as much as 50%.

For more information on how to select the right block storage for your workload and to see how customers have benefitted from Hyperdisk, read this blog.

3. Consider custom compute classes

To get the most out of our latest-generation VMs, Google Kubernetes Engine (GKE) custom compute classes (CCC) offer an advanced way to optimize compute choices and provide high availability. Instead of being limited to a single machine type for your workloads, you can define a prioritized list of VM instance types. This allows you to set the newest, most price-performant VMs — including our latest-generation VMs — as your top priority. GKE custom compute classes provide the capability to automatically and seamlessly spin up instances based on your specified priority list. This feature helps you maximize the availability of your compute capacity while still aiming for the most cost-effective options, so your workloads can scale reliably without manual intervention.

Here are some specific use cases for how custom compute classes can help you optimize costs:

• Autoscaling cost-performant fallbacks: When demand peaks, you might be tempted to autoscale using a highly available but less cost-efficient VM type. CCC allows you to take a tiered approach. You can set up several cost-efficient fallback alternatives, so that as demand increases, GKE first attempts to use the most cost-effective options, and progressively moves to the other choices in your list when necessary to meet demand.
• AI/ML inference: Running AI/ML inference workloads often involves significant compute resources. Instead of maintaining a large, static reservation that might sit idle during off-peak times, CCC lets you provision a minimal base reservation and leverage more cost-effective capacity types, such as Spot VMs, to handle peak inference demand — all orchestrated through your CCC configuration.
• Adopting new VM generations: Combine the power of GKE custom compute classes with Compute Flexible committed use discounts (Flex CUDs) to de-risk the adoption of new, cost-efficient VM series like N4 and C4. With CCC, you can define fallback options, providing workload resilience, while Flex CUDs offer financial adaptability, as the discounts apply across your total eligible compute spend, regardless of the specific VM series you use. This dual approach is a safe, cost-effective strategy for leveraging the latest hardware without disruption. For more information, read this blog.
• Using flexible Spot VMs: Spot VMs offer significant savings but can be preempted. Being constrained to a single Spot VM shape increases the risk that capacity will not be available. With CCC, you can define multiple fallback Spot VM types. This “spot surfing” capability allows the application to remain on cost-efficient Spot capacity by automatically pivoting to alternative Spot instance types if the primary choice is unavailable.

In short, by leveraging GKE CCC, you can artfully mix and match various VM types and consumption models, including On-Demand, Spot, DWS FlexStart, and instances covered by CUDs, to build a resilient and highly cost-optimized infrastructure that adapts to the unique needs and patterns of your workloads.

4. Leverage custom machine types (CMT)

Custom machine types, available on N4 VMs, allow you to precisely configure virtual machines to your exact specifications. Rather than selecting from predefined machine types that might include excess capacity, you can tailor the CPU-to-memory ratio specifically for your workloads, so you only pay for resources you actually use. This targeted approach minimizes waste and can significantly reduce your cloud spend, especially when migrating from on-premises to Google Cloud or from other cloud providers.

This flexibility becomes particularly valuable if your applications have unique resource profiles that don’t align well with our standard offerings. Custom machine types let you create the perfect environment for your needs. By avoiding the compromise of over-provisioning certain resources while potentially constraining others, you can achieve both better performance and more efficient spending across your Compute Engine deployment.

As an example, take a memory-intensive workload that runs best with 16 vCPU, and 70 GB memory. Normally, you would need to pick a VM with 128 GB memory with our standard shapes, or in other cloud contexts, resulting in higher costs to run your workload due to the extra provisioned resources. Instead, with custom machine types, you can easily launch a VM with 16 vCPU and 70 GB memory, resulting in an 18% cost savings vs standard N4-highmem-16 VMs.

5. Make the most of committed use discounts

CUDs are a strategic cost-saving opportunity for organizations with steady, predictable computing needs. By committing to resource usage over one- or three-year periods, you can reduce cloud costs by up to 70% compared to on-demand pricing. This approach not only helps ensure budget predictability but also converts fixed infrastructure spending into a financial advantage, making it ideal for stable workloads that support core business functions.

Google Cloud offers flexible CUD structures to align with various operational models. Resource-based commitments target specific machine types and regions, flexible commitments apply discounts across projects, regions, and machine series — great for dynamic environments. By analyzing historical usage and forecasting future needs, you can identify workloads suited for these discounts, reinvesting the savings into innovation and scaling initiatives.

6. Manage unused disk space

You pay for the total provisioned disk space, regardless of how much you actually use. Many organizations tend to over-provision storage “just in case,” which often leads to unnecessary and costly waste. For instance, if you provision a 100GB disk but only use 20GB, you’re still paying for the entire 100GB. Being intentional and precise with your storage allocations — rather than rounding up to common sizes — can lead to significant cost savings.

To optimize spending, it’s important to adopt a few best practices. Using Ops Agent, regularly audit disk usage across your infrastructure to identify and eliminate inefficiencies. Resize disks to align with actual consumption, allowing a reasonable buffer for growth. Implement automated alerts in Cloud Monitoring to detect underutilized disks and take corrective action. For stateless applications, consider using smaller boot disk images to minimize overhead and reduce costs even further.

In addition, consider the following optimization strategies to further reduce costs and improve efficiency:

• Use Google Cloud’s monitoring tools to track CPU, memory, and disk usage over time.
• Establish a regular review cycle to identify and right-size over-provisioned resources.
• Test workloads across different VM configurations to find the optimal balance between cost and performance.

7. Use Spot VMs

Spot VMs provide the same machine types and configuration​​ options as standard virtual machines but at a significantly reduced cost — typically offering a 60% to 91% discount. This cost efficiency comes with the tradeoff of potential preemption at short notice, making them most suitable for workloads that are fault-tolerant and can recover quickly from unexpected interruptions. Spot VMs are designed to take advantage of unused compute capacity, allowing you to optimize your cloud spending without compromising access to high-performance resources.

Strong use cases for Spot VMs include batch processing jobs, big data and analytics workloads, continuous integration and deployment (CI/CD) pipelines, stateless web servers running in autoscaling groups, and compute-heavy tasks. When properly architected to handle interruptions — for example, by using job checkpointing, load balancing, task queues, or via GKE custom compute classes (see more above) — Spot VMs can play a critical role in minimizing infrastructure costs while maintaining high availability and system resilience. Leveraging Spot VMs in these scenarios lets you scale cost-effectively, especially when compute demand is variable or time-flexible.

8. Use optimization recommendations

Google Cloud’s Recommenders are a powerful tool designed to help you optimize your cloud resources efficiently. When browsing the Google Cloud console, you may see lightbulb icons next to specific resources — these indicate potential improvements identified by Google’s recommendation engine. By analyzing real-time usage patterns and current resource configurations, the Recommender delivers actionable insights tailored to each user’s unique environment. This intelligent system highlights opportunities not only to reduce costs but also to enhance security, performance, reliability, management efficiency, and environmental sustainability.

For example, there are idle VM recommendations to help you identify VM instances that have not been used over the last 1 to 14 days. Common recommendations include switching to more suitable machine types, rightsizing underutilized compute instances, or adopting more cost-effective storage solutions. The tool allows you to apply many of these changes directly, streamlining the optimization process. By continuously evaluating workloads and offering these automated, data-driven suggestions, the Recommendation Hub helps organizations maintain cloud performance while managing costs more effectively.

9. Take advantage of auto-scaling and scheduling

Matching your compute resources to actual demand patterns is one of the most effective ways to reduce cloud waste and improve overall cost efficiency. Many organizations over-provision their resources to handle peak workloads, leaving machines underutilized during off-peak periods. By aligning compute capacity more closely with real-time or predictable usage patterns, such as business hours or seasonal trends, you can significantly cut unnecessary spending without sacrificing performance.

Autoscaling is the key to achieving this efficiency. In fact, customers who leverage Google Compute Engine autoscaling for their virtual machines have seen average infrastructure cost savings of more than 40%.

You can implement autoscaling strategies to dynamically adjust resources based on CPU utilization, load balancing capacity, or custom application metrics, so that workloads receive the necessary compute power when needed, while scaling down automatically during low-demand periods.

For workloads with predictable patterns, such as those that fluctuate with business hours or planned seasonal events, schedule-based scaling is a particularly powerful tool. This approach allows you to proactively increase resources in anticipation of high demand and scale them down during lulls, for the performance you need without constant over-provisioning.

In addition to autoscaling, several practical implementation techniques can further optimize your resource usage. Setting up instance scheduling lets you automatically start and stop development and test environments according to business hours — a simple yet highly effective approach that can lead to cost savings of up to 70%. You can also leverage maintenance windows to reduce disruptions and resource consumption, by concentrating updates and system changes into low-usage periods. Together, these tactics help maintain high availability and performance while keeping infrastructure costs under control.

10. Understand your spend with detailed billing analysis

Before implementing any cost-saving strategies in Google Cloud, it’s essential to understand your current spending in detail. Google Cloud’s billing panel offers granular visibility into your expenses, including costs broken down by individual SKUs. This level of transparency lets you track where your money is going and identify potential inefficiencies. Begin by regularly reviewing your billing dashboard to monitor usage trends and spot anomalies. Applying labels and tags to your resources can further help categorize and attribute costs accurately, especially in complex environments with multiple projects or departments.

In addition, setting up budget alerts is a practical way to stay ahead of overspending by notifying you when costs approach or exceed predefined thresholds. It’s also important to identify and eliminate unused or idle resources, such as virtual machines or persistent disks that are no longer in active use — these can often be shut down or deleted to immediately reduce costs. By thoroughly analyzing your cost structure, you can uncover “low-hanging fruit” — resources that provide little or no value — and make data-driven decisions to optimize your cloud usage efficiently.

11. Consider serverless alternatives

Last but not least, Google Cloud’s serverless computing offerings provide a compelling alternative to traditional virtual machines, can deliver better cost efficiency, simplified operations, and greater scalability. By abstracting away infrastructure management, serverless platforms allow teams to focus on writing and deploying code without worrying about provisioning, scaling, or maintaining servers. This shift can not only reduce operational overhead but also cut costs by aligning compute spending directly with application usage.

There are multiple serverless options available, each tailored to different workloads. Cloud Run is designed for running containerized applications that need rapid scaling and flexible deployment. Cloud Run Functions supports lightweight, event-driven code execution for microservices or automation tasks. GKE (Autopilot Mode) simplifies Kubernetes operations by automatically managing nodes and scaling, allowing you to run Kubernetes workloads without handling the underlying infrastructure. All these options charge based on usage not allocation, significantly reducing costs associated with idle resources and over-provisioning. This makes them especially beneficial for variable or unpredictable workloads. Cloud Run and GKE both support GPU’s and flexibility to move between the two. You can start with Cloud Run then move to GKE or vice-versa. Some customers also leverage both offerings for workloads. The rule of thumb is to start with GKE if you need access to the Kubernetes API. Otherwise, start with Cloud Run.

Start reducing your costs today

Migrate to Google Cloud and optimize your infrastructure costs without compromising on what your workloads need. If you are new to Google Cloud, start with a migration assessment. Google Cloud’s Migration Center can help you with a clear understanding of your potential savings by migrating to Google Cloud, with detailed recommended paths for your workloads, along with TCO reports. Apply the strategies in this article and unlock substantial cost savings.

Building the best AI applications requires both the freedom to choose the most powerful, specialized model for the task at hand, and a platform that can handle them all. This flexibility is core to the Vertex AI platform, and today, we’re taking a significant step forward in our commitment to giving you unparalleled choice and control.

We are excited to announce that you can now securely deploy a growing selection of leading proprietary models from industry partners, including AI21 Labs, CAMB.AI, CSM, Mistral AI, Qodo, and Virtue AI, with models from Contextual AI and WRITER coming soon. You can deploy these models — including closed-source models and those with restricted commercial licenses — directly into your own Virtual Private Cloud (VPC).

You will find all of these models in the Vertex AI Model Garden, our central gateway to over 200 foundation models, including Google’s versatile Gemini family, leading open models, and third-party models. We provide a single, curated catalog where you can discover, test, and deploy the ideal model for your application.

Announcing self-deployable proprietary models in your VPC

For organizations that require maximum control over their data and infrastructure, you can now self-deploy powerful proprietary models from leading AI model builders directly within your VPC. With this new capability, you can acquire commercial licenses via Google Cloud Marketplace and deploy models securely within your environment, all while meeting Google Cloud’s high standards for security and compliance. Self-deploying proprietary models with Google Cloud gives you a number of benefits:

• Deploy models within your VPC with full adherence to your VPC-SC policies, providing the highest assurance that your proprietary business data never leaves your environment. You can evaluate and deploy third-party models to production on a trusted platform.
• You can optimize for performance or cost by selecting from a range of available machine types. Scale your replica count up or down manually to meet workload demands, or configure auto-scaling policies for hands-free management. Deploy to specific Google Cloud regions of your choice to achieve data compliance in your target markets or select locations for low latency delivery to your customers.
• Discover, license, and deploy a curated selection of proprietary models from industry-leading providers, all in one place. We’re launching with models from eight partners—AI21 Labs, CAMB.AI, Contextual AI, CSM, Mistral AI, Qodo, Virtue AI, and WRITER. These models cover a wide range of use-cases and specializations. This is just the beginning, and you’ll see us continue to expand our catalog with the latest generative AI models.
• Go from model discovery to production with ease. You can procure commercial licenses and deploy the models with just a few clicks directly from the Model Garden console. Our fully managed AI inference service handles the underlying infrastructure, so you’re free to focus on building your application.
• Get started with simple pay-as-you-go pricing, so you only pay for what you use. You control your costs by scaling your deployment to meet your needs, and not deal with artificial limits or quota caps. You can further optimize costs by applying your existing Google Cloud committed-use discounts (CUDs) or reservations.

Meet Our Launch Models

Explore the new models available today for self-deployment in your VPC:

• AI21 Labs – Jamba Large 1.6: Delivers leading model quality at fast speed, making it an excellent choice for private enterprise deployment.
• CAMB.AI – MARS7: Enables you to ship production-ready voice applications with hyper-realistic, multilingual text-to-speech (TTS) outputs featuring optional voice cloning and fine-grained emotional control.
• (Coming soon!) Contextual AI – Reranker: Designed to significantly enhance the relevance and quality of Retrieval-Augmented Generation (RAG) systems.
• CSM – Cube: A generative AI model that transforms 2D images into detailed 3D models with remarkable precision and efficiency.
• Mistral AI – Codestral (25.01): Explicitly designed for code generation tasks, helping developers write and interact with code through a shared instruction.
• Qodo – Embed-1: A suite of large-scale code embedding models that enhance search accuracy for RAG by enabling efficient code and text retrieval.
• Virtue AI – VirtueGuard: An enterprise-ready AI guardrail model that enables real-time content security, policy enforcement, and regulatory compliance with multilingual support for generative AI systems.
• (Coming soon!) WRITER – Palmyra X4: Enterprise-grade LLM that combines a 128K token context window with a suite of capabilities, including advanced reasoning, tool calling, LLM delegation, built-in RAG, code generation, structured outputs, multi-modality, and multilingual

How to Get Started

You can deploy these new models in three simple steps:

1. Visit the Vertex AI Model Garden. On the left-hand navigation tab, under “Model Collections,” select “Self-deploy partner models”.
2. Select the model from a partner that you choose to deploy. To use the selected model, purchase a license by clicking “Enable”.
3. Your license is active in a few seconds. Simply click “Deploy” to configure and deploy the model endpoint in your VPC using Model Garden’s one-click deployment workflow.

We are committed to providing the most open and flexible AI platform for the enterprise. With even more choice and the fine-grained control and security of your own environment, you have everything you need to innovate responsibly. Explore the new models in Model Garden today and start building today!

Many data science teams rely on Apache Spark running on Dataproc managed clusters for powerful, large-scale data preparation. As these teams look to connect their data pipelines directly to machine learning models, there’s a clear opportunity to simplify the integration. But running inference on a Spark DataFrame using a model from Vertex AI typically requires custom development, making it complex to build a single, end-to-end workflow.

To solve this problem, we are developing a new open-source Python library designed to simplify AI/ML inference for Dataproc. This library connects your Apache Spark jobs to use popular ML frameworks and Vertex AI features, starting with model inference. Because the library is open-sourced, you will be able to use it directly in your application code with full transparency into its operation.

How it works

Dataproc ML is built to feel familiar to Spark users, following a SparkML-style builder pattern. You configure the model you want to use, and then call .transform() on your DataFrame. Let’s look at a few common inference use cases.

1. Apply Gemini models to your Spark data

You can apply generative AI models, like Gemini, to columns in your Spark DataFrame. This is useful for tasks like classification, extraction, or summarization at scale. In this example, we take a DataFrame with “city” & “country” columns and use Gemini to create a new column by providing a simple prompt.

You can test in your local environment by installing from PyPi:

To deploy/test at scale, create a Dataproc version 2.3-ml cluster:

Copy this example to a file gemini_spark.py.

The handler is flexible to support customized options, as explained in the documentation.

Submit this job to your Dataproc cluster:

2. Run inference with PyTorch and TensorFlow models

In addition to calling Gemini endpoints, the library also allows you to run inference with model files loaded directly from Google Cloud Storage. You can use the PyTorchModelHandler (and a similar handler for TensorFlow) to load your model weights, define a pre-processor, and run inference directly on your worker nodes. This is useful when you want to run batch inference at scale without managing a separate model serving endpoint.

Built for performance

This library isn’t just a simple wrapper. It’s designed for running inference on large Dataproc clusters and includes several optimizations for inference:

• Vectorized data transfer: We use pandas_udf to efficiently move data between Spark and the Python worker processes.
• Connection re-use: Connections to the endpoint are re-used across partitions to reduce overhead.
• Retry logic: The library automatically handles errors like HTTP 429 (resource exhausted) with exponential backoff and retries.

Get started

You can start using it today by checking out the open-source repository and reading our documentation.

Looking ahead, we plan to add the following features to this library in the coming months.

1. Spark Connect support: This would also allow using above functionalities within BigQuery Studio notebooks.
2. Vertex AI integrations: To ease inference, we plan to add more ModelHandlers to:
   1. Directly call a vertex model endpoint for online inference
   2. Refer to Vertex models and localize them to Spark workers
   3. Refer to models hosted in Vertex Model Garden including embedding models
3. More Optimizations: Auto-repartition input dataframes to enhance inference runtime
4. Third-party integrations: Refer to open sourced models in HuggingFace

We are actively working on including this library by default in Dataproc on Google Compute Engine ML images and Google Cloud Serverless for Apache Spark runtimes.

We look forward to seeing what you build! Have feedback or feature requests to further simplify your AI/ML experience on spark? Reach us at dataproc-feedback@google.com.

In today’s complex and ever-evolving threat landscape, federal agencies require secure, reliable, and innovative solutions to fulfill their critical missions. Google Pixel phones have been added to the Department of Defense Information Network (DoDIN) Approved Products List (APL) – a significant milestone that underscores our commitment to providing federal agencies with secure, cutting-edge technology. Combined with Google’s comprehensive cloud ecosystem, federal agencies are further empowered to enhance security, enable better communication, improve efficiencies, and streamline operations. Achieving placement on the (DoDIN) Approved Products List (APL) establishes a crucial foundation of trust and enables government agencies to confidently procure Google Pixel phones, knowing they adhere to the highest security standards required.

The Pixel advantage

Google Pixel phones are built on a foundation of mission-ready resilience and intelligent security that is seamlessly integrated into the Google ecosystem. Pixel 9 phones are the highest rated for security features¹ empowering government employees to securely connect and collaborate from virtually anywhere—even the most remote environments. Leveraging 5G’s mobile cellular technology—which provides high speeds, low latency, and increased mobility—equips military operations with new capabilities that are essential for reliable high-priority communication.

To see these capabilities in action, consider how one federal agency is modernizing its global operations.

A federal agency procuring energy resources for the military sought to improve fuel inspection and auditing capabilities across their 500+ global sites. Current processes are manual and data is physically archived, leading to issues like data fidelity and real-time access problems. They will be using Google products, including Pixel phones for on-site data capture and secure transmission, to digitize and standardize data collection that will then feed into a centralized system to predict compliance and environmental issues.

Building the future of the Public Sector

To support the critical work of federal agencies, Google Cloud offers a comprehensive technology ecosystem with integrated AI, analytics, and robust security. By combining Pixel phones and Google Cloud services, agencies gain a complete, unified tech stack for secure connectivity and collaboration, enabling them to fully leverage Google’s innovative technologies.

The power of combining secure, on-device data capture with cloud-based AI is also transforming how essential public services are delivered. For example, Google Public Sector in partnership with MTA, piloted TrackInspect, an AI-powered system that uses Pixel smartphones to detect potential subway track defects. By analyzing real-time sensor data, the system identified 92% of defect locations found by human inspectors, enabling proactive maintenance and reducing service disruptions. This innovative approach combines cloud technology, AI, and generative AI to improve track safety and efficiency. The pilot program demonstrates how technology can enhance infrastructure maintenance, resulting in smoother and more reliable service for millions of daily riders.

Discover how Google Public Sector can help empower your agency

Attaining DoDIN APL certification is just one example of why Pixel’s seamless Google Cloud integration and robust security features make it a strong choice for public sector applications. Check out our Google Pixel Public Sector 1-pager for more insight into how Pixel DoDIN APL certified phones can help your agency streamline communications, enhance field productivity, and improve collaboration. With built-in security, on-device AI, and secure remote management features, Pixel phones are designed to meet strict government security requirements. Explore Pixel phones firsthand and register for the Google Public Sector Summit on October 29th in Washington D.C. to learn more.

Footnote:

1. Based on a third-party global research firm report. Evaluation considered features that may not be available in all countries. See goo.gle/mobilesecurity for more information.

It’s been exciting to see the capabilities of Nano Banana, our latest image editing model available in Gemini 2.5 Flash Image, go viral. And with transformative workflows like these, it is easy to see why:

The incredible response makes it clear: enterprises now have the ability to quickly create and refine high-quality media across more formats and channels than ever before, without compromising on appeal, consistency, or security.

That’s why we’re thrilled to announce major updates across our suite of generative media models—including Gemini 2.5 Flash Image (now GA!), Veo, Imagen, and Gemini 2.5 Text-To-Speech — on Vertex AI. These updates help you create faster, with more control, and across all the formats that matter most: sight, sound, and motion. Let’s take a look.

Gemini 2.5 Flash Image is Generally Available (GA) on Vertex AI

We are excited to announce the General Availability of Gemini 2.5 Flash Image. Our state-of-the-art image generation and editing model is now production ready and backed by Google Cloud’s enterprise-grade infrastructure and security. In addition, the model now creates images across multiple aspect ratios and supports batch processing.

We’re already seeing incredible adoption of Gemini 2.5 Flash Image. Here’s an example of how companies are putting the pushing the creative boundaries of Gemini 2.5 Flash Image:

“Gemini 2.5 Flash Image and such high-quality AI tools mean, quite literally, nothing is off limits anymore. As a result, our team has never been more creative. We’re collaborating on ideas, able to visualize them faster and launch campaigns in days, instead of weeks. Our core mission is always to give creators and businesses the most advanced AI tools, and putting Gemini 2.5 Flash Image in their hands fulfills that promise. This is, without a doubt, the most exciting time to be a creator.” – Shahar Aizenberg, CMO, Artlist.io

“Gemini 2.5 Flash Image has redefined what’s possible for Mercado Libre’s Photo Studio. The model’s creativity, aesthetic quality, and precise instruction following have elevated our product listings and unlocked new possibilities. Today, our only limit is imagination.” – Franco Seia, Software Development Manager, Mercado Libre

Veo: Dream up creations in new formats with greater control

Veo 3, our latest video generation model with native audio and dialogue, has been rapidly adopted by creators looking to bring their stories to life with unprecedented control. We’ve been listening to your feedback, and we’re excited to announce new features that make Veo 3 even more powerful on Vertex AI:

• Output vertical formats for social media: You told us you wanted to create vertical videos, and we listened! Veo 3 and Veo 3 Fast now support a 9:16 aspect ratio. Creators can produce larger, more immersive visuals that are optimized for the vertical orientation of social media platforms. Say goodbye to awkward cropping!
• Control the flow and timing of your story: With the 4, 6, or 8 second duration options, you can seamlessly tune your video content for transitions and intercut scenes, allowing you to craft more flexible narratives for your needs.

Our customers are already leveraging Veo to transform their creative workflows and connect with their audiences in powerful new ways.

“For Palo Alto Networks “Be a Genius. Deploy Bravely” campaign, we proved you no longer have to choose between speed, creativity, and cost. With Gemini and Veo 3, you get all three. Like our customers, we are navigating the incredible promise of the AI revolution. The smartest move in this new era isn’t just to adopt AI, but to do it securely.” – Kelly Waldher, CMO, Palo Alto Networks

“We’ve always believed that the future of creativity is a dynamic partnership between creatives and technology. That’s why we’re bringing together the best-in-class AI models across video, images and audio directly into our unlimited Envato subscription. Early signals indicate strong usage of models like Veo 3, accelerating our community’s creativity and empowering them to thrive in the process.” – Hichame Assi, CEO, Envato

Imagen 4 is Generally Available (GA) on Vertex AI

Our leading text-to-image model, Imagen 4, is engineered for creativity and speed. It delivers photorealistic images, sharp clarity, and text rendering and typography, bringing your imagination to life faster than ever before. It is Generally Available and production ready on Vertex AI.

Shutterstock, a family of brands delivering scalable creative and GenAI solutions, is using Google’s Imagen 4 models to power high-quality, commercially ready AI images.

“At Shutterstock, our mission is to empower businesses with the essential, universal ingredients to make their work more effective. By bringing Google’s Imagen 4 models into our AI Image Generator, we’re making it easier than ever for teams to go from an idea to impact with market-ready visuals in seconds. Imagen 4 allows us to deliver high-quality, commercially ready outputs that meet the standards our customers expect. This integration ensures Shutterstock customers are always equipped with the most advanced, future-ready tools in creative AI.” – Keenan Kadam, Senior Product Manager, Shutterstock

Gemini 2.5 TTS (Text to Speech) is Generally Available (GA) on Vertex AI

Our powerful generative media capabilities also include audio. Leverage Gemini 2.5 Text-to-Speech for the creation of high-fidelity voice applications, all with the security and scale of Vertex AI.

We are thrilled to announce that Gemini 2.5 Text-to-Speech (TTS) is now Generally Available across both Pro and Flash models. This model prioritizes human-like expression and control, transforming how you build voice applications:

• Studio-Quality Dialogue, Now GA: Stop relying on choppy, single-speaker systems. Gemini 2.5 Flash and Gemini 2.5 Pro TTS are now ready for production, enabling you to generate dynamic, multi-speaker dialogue in a single API call for podcasts, audiobooks, and rich conversational customer service.
• Advanced Style and Tone Control: Leverage natural language prompts to dictate the performance, not just the text. You can now control the voice’s tone, emotional expression, and accent, ensuring your brand’s voice is delivered with perfect fidelity.
• Global Reach, Perfect Delivery: Gemini 2.5 Flash and Gemini 2.5 Pro TTS are now available across more than 70 languages, guaranteeing your global audience receives the same high-quality, expressive voice experience, regardless of region.

Which gen media model to use, and when

We have a suite of options for enterprise-grade work because we know choice is important when it comes to the right model for your project. If you’re not sure where to start, here’s a quick cheat sheet:

1. Choose Veo 3 if your workflow demands dynamic, high-quality video creation with granular control over scenes, characters, and narrative flow. It’s perfect for social media content, marketing campaigns, and any project where bringing stories to life through motion is key. Your input can be text, images, or a combination. For the latest Veo 3 pricing go here.
2. Choose Gemini 2.5 Flash Image as a starting point for image creation or if your workflow is iterative and requires creating or editing an image with strong visual consistency. It’s the right choice for conversational editing, sketch-to-image tasks, style transfers, and adapting existing visuals. Your input is often a combination of images and text prompts.
3. Choose Imagen 4 if your workflow is focused on generating net-new images from text with speed and higher resolution. It’s built for high-volume text-to-image applications where speed and resolution are your primary concerns.
4. Choose Gemini 2.5 Flash or Gemini 2.5 Pro TTS (Text-to-Speech) if your workflow is centered on bringing text to life with high-quality, emotionally expressive audio. It’s the right choice for creating lifelike voice agents, professional narration for content like podcasts and e-learning including multi-speaker synthesis, and dynamic character voices for gaming and entertainment. Your input is text.

Get started with enterprise-grade creativity on Vertex AI

Gemini 2.5 Flash Image, Veo 3, Imagen 4, and Gemini 2.5 TTS are available on Vertex AI today.

Dive into the Vertex AI Studio to get started with Gemini 2.5 Flash Image and Gemini 2.5 TTS today. For Veo 3 and Imagen 4, get started at Vertex AI Media Studio.

Technical foundation: security layers

The oracle system architecture addresses distinct threat models, each requiring different technical countermeasures:

Adding features to an app can be hard. One minute you’re writing code, the next you’re switching to the PostgreSQL database client to run a query, and then it’s over to the console to check on your instances. For example, let’s say you wanted to add search capabilities. This can mean adding the right extensions to your PostgreSQL database and learning how to use it. How do you know which extension to use and how to use it in your queries? What if you could do it all from one place, using plain English?

The recently announced Gemini CLI extension for PostgreSQL is here to do just that. It’s a powerful tool that brings the magic of natural language and the convenience of the command line to your database tasks.

Let’s see it in action: The fuzzy search challenge

Imagine you want to add a “fuzzy search” feature to your app — you know, so users can find a “t-shirt” even if they type “tshirt”.

Normally, this would involve a bunch of research and manual steps. But with the Gemini CLI, it’s a conversation:

1. The ask: You tell Gemini CLI you need to implement fuzzy search.
2. The smart suggestion: Gemini CLI immediately identifies that the pg_trgm extension is the perfect tool for the job.

The proactive check: It doesn’t just assume you have the extension. It runs list_installed_extensions() to see if pg_trgm is already installed. If not, it installs it for you!

1. The performance tip: To make your search blazing fast, it then recommends creating a GIST or GIN index on the relevant table column (e.g., the details column in your product_inventory table).

The code: Finally, it provides you with a sample query, showing you exactly how to execute the fuzzy search.

In just a few steps, you’ve gone from a feature request to a fully implemented, optimized solution. That’s not just a command line; that’s a database assistant.

Key features you’ll love

To recap, the Gemini CLI extension for PostgreSQL lets you:

• Talk to your database: Use natural language to get what you need, from simple schema exploration to complex tasks like setting up extensions.
• Exert full lifecycle control: From creating a new PostgreSQL instance to managing users and permissions, this extension has you covered.
• Generate code on the fly: Speed up your development by automatically generating data classes and other code snippets based on your table schemas.

To get started, use the Gemini CLI extension for PostgreSQL.

For AI agents to be really useful, they need to be able to securely interact with enterprise data. In July, we introduced a toolset to help AI agents interact with and analyze business data in BigQuery through natural language, and with just a few lines of code. Today, we’re taking the next step, with “Ask data insights” for Conversational Analytics and the “BigQuery Forecast” for time-series predictions, going beyond fetching metadata and executing queries to full-scale data analysis and predictions. Both tools are available today in the MCP Toolbox as well as Agent Development Kit’s built-in toolset.

Let’s dive into what you can do with these new tools.

ask_data_insights: Converse with BigQuery

With the ask_data_insights tool, you can now answer complex questions of your structured data in BigQuery using plain English.

Built on top of the powerful Conversational Analytics API, ask_data_insights enables an agent to utilize the API to offload the task of understanding the user’s question, pulling in relevant context, formulating and executing the queries, and summarizing the answer in plain English. Along the way, the ask_data_insights tool shows its work, returning a detailed step-by-step log of its process, so you have full transparency into how it arrived at the answer.

Predict the future with BigQuery Forecast

Information without insights is just noise. The ability to predict future trends, whether sales, user traffic, or inventory needs, is critical for any business. BigQuery Forecast simplifies time-series forecasting using BigQuery ML’s AI.FORECAST function based on the built-in TimesFM model.

With BigQuery Forecast, the agent can run the forecasting job directly within BigQuery, without you having to set up machine learning infrastructure. Point the tool at your data, specify what you want to predict and a time horizon, and the agent will make its predictions using TimesFM.

New tools in action: Building a Google Analytics Data Agent

Let’s explore how to build a simple agent to answer questions about Google Analytics 360 data using ask_data_insights and BigQuery Forecast. For this demo,

• The data is stored in BigQuery tables. Users of this agent only require read access to these tables, which are available under the BigQuery public dataset. bigquery-public-data.google_analytics_sample.
• We will use ADK to build this agent and “adk web” to test it.
• We are using one tool from the ADK’s built-in tools and one from the MCP toolbox. You can choose to use either option depending on your agent architecture and needs.

This diagram shows the architecture of this simple agent:

And here is the agent code:

Using the agent code above, let’s turn to the ADK’s developer UI, i.e., adk web, to test the agent and see it in action.

First, let’s use the tools to understand our data…

Then, let’s see if the agent can answer a business question.

As you can see above, the Conversational Analytics API is equipped with the ability to perform deep thinking, so it can provide rich insights into our question.

Now, let’s see if the agent can predict the future.

Short answer, yes, yes it can, with a 95% confidence level. With these tools, the power of the TimesFM model is finally available to business users, regardless of their technical skill level.

Bring analysis and forecasts to your data

These new BigQuery capabilities will help developers reimagine how they build data-driven applications and agents. Together, we believe the combination of AI-powered Conversational Analytics and powerful, built-in forecasting capabilities will make performing sophisticated data analysis easier than ever.

Learn more about the ask_data_insights and BigQuery Forecast tools in the MCP Toolbox for databases and the core Agent Development Kit.

Today, we’re announcing the general availability of Claude Sonnet 4.5, Anthropic’s most intelligent model and its best-performing model for complex agents, coding, and computer use, on Vertex AI.

Claude Sonnet 4.5 is built to work independently for hours, maintaining clarity while orchestrating tools and coordinating multiple agents to solve complex problems. It’s designed to excel at long-running tasks with enhanced domain knowledge in coding, finance, research, and cybersecurity. Key use cases include:

• Coding: Autonomously complete long-horizon coding tasks. Plan and execute software projects that span hours or days, as well as everyday development tasks.
• Cybersecurity: Deploy agents that autonomously patch vulnerabilities before exploitation, shifting from reactive detection to proactive defense.
• Financial analysis: Handle everything from entry-level financial analysis to advanced predictive analysis like continuously monitoring global regulatory changes and preemptively adapting compliance systems.
• Research: Handle tools, context, and deliver ready-to-go office files to drive expert analysis into final deliverables and actionable insights.

We’re also announcing Vertex AI support for Anthropic’s upgrades to Claude Code—including a VS Code extension and the next version (2.0) of the terminal interface, complete with checkpoints for more autonomous operation. Powered by Claude Sonnet 4.5, Claude Code now handles longer, more complex development tasks than the previous version.

Get started

Start building with Claude Sonnet 4.5 today by following these instructions:

1. Navigate to the Claude Sonnet 4.5 model card in the Vertex AI Model Garden, select “Enable”, and follow the proceeding instructions. You can also find and easily procure Claude Sonnet 4.5 on Google Cloud Marketplace.
2. Explore our sample notebook for help getting started.
3. Visit our documentation for pricing and regional support details.

The Vertex AI advantage: production-ready agents and applications

Bringing powerful models like Claude Sonnet 4.5 to production quickly requires more than just API access; it requires a unified AI platform. Vertex AI gives you an all-in-one platform to build, deploy, and manage your AI with confidence:

• Orchestrate multi-agent systems: Build agents with a flexible, open approach through Agent Development Kit (ADK), then scale them in production with Agent Engine.
• Get committed capacity: Reserve dedicated capacity and prioritized processing for your Claude Sonnet 4.5 applications at a fixed cost. This is made possible with provisioned throughput. To get started, contact your Google Cloud sales representative.
• Boost model performance and efficiency: Get the most out of Claude with supported features. Run large-scale jobs with batch predictions, analyze codebases with a 1 million token context window, reduce costs with prompt caching, and ground responses with citation support. For detailed information on Claude-supported features, refer to our documentation.
• Deploy with highly efficient infrastructure: Deploying on Vertex AI means running on infrastructure that is purpose-built for AI and designed to provide optimal performance and cost across your workloads. The global endpoint for Claude also enhances availability and reduces latency by dynamically serving traffic from the nearest available region.

Operate securely, by design: Build securely from day one, knowing your data is protected and you can easily manage compliance. This is provided by Vertex AI and Google Cloud’s robust, built-in security and data governance controls.

How customers are building with Claude on Vertex AI

Leading organizations are already leveraging the powerful combination of Claude and Google Cloud to drive significant business impact.

Augment Code is powering its AI coding assistant, which helps developers navigate and contribute to production-grade codebases, with Claude on Vertex AI.

“What we’re able to get out of Claude is truly extraordinary, but all of the work we’ve done to deliver knowledge of customer code, used in conjunction with Claude and the other models we host on Google Cloud, is what makes our product so powerful.” – Scott Dietzen, CEO, Augment Code

spring.new is helping users build custom business applications and tools in hours using natural language prompts.

“Customers tell us that our platform, powered by Claude models and Google Cloud, enables one person to create applications in one to two hours that previously took up to three months.” – Amitay Gilboa, CEO, spring.new

TELUS built its generative AI platform, Fuel iX™, on Google Cloud to give its team members a choice of curated AI models, like Claude, inspiring engineering excellence and enterprise-wide productivity.

“Getting a model as powerful as Claude on Vertex AI is a win-win that makes life so much easier. We get a model that excels at tool calling on a comprehensive platform that integrates with our core Google Cloud workloads like GKE and Cloud Run — that’s the magic.” – Justin Watts, Distinguished Engineer, TELUS

Welcome to the second Cloud CISO Perspectives for September 2025. Today, Google Cloud COO Francis deSouza offers his insights on how boards of directors and CISOs can thrive with a good working relationship, adapted from a recent episode of the Cyber Savvy Boardroom podcast.

As with all Cloud CISO Perspectives, the contents of this newsletter are posted to the Google Cloud blog. If you’re reading this on the website and you’d like to receive the email version, you can subscribe here.

Boards should be ‘bilingual’ in AI and security to gain a competitive advantage

By Francis deSouza, chief operating officer, Google Cloud

By learning the language of AI for defense, boards can be better prepared to use AI to create a competitive advantage.

Cybersecurity should be a core duty of every board member, not just those serving on audit and risk committees. Becoming bilingual in AI can help board members focus on why they should understand their organization’s security posture, and be prepared for potential breaches. But there’s much more that boards can do — here are four steps leaders can take to drive effective change in today’s dynamic environment.

1. Integrate cybersecurity into business strategy

What used to be a landscape dominated by individual hackers has now dramatically expanded to sophisticated groups that have been specifically formed to extract value from organizations by stealing and ransoming their data.

While it’s important to be fluent in business strategy, boards should also work with security leaders towards integrating cybersecurity into their overall roadmap. Boards can encourage a collaborative approach to align cybersecurity with critical business services, which can help strengthen security posture, protect critical assets, and enhance resilience against evolving and emerging threats.

2. Develop a framework for cybersecurity investments

Boards should ask questions to ensure cybersecurity investments deliver real business value — beyond compliance. Key areas for boards to investigate include identifying and understanding the protection of critical digital and physical assets with software components, assessing the maturity level of protection, and knowing the potential cost of different types of breaches.

Here’s where boards should encourage third-party assessments, running simulations, and tabletop exercises to help prepare an organization for breach responses. It’s also important for boards to develop a framework for cybersecurity investments to help them benchmark spending against industry data, and assess the effectiveness of that investment.

When boards understand the risks and costs associated with different types of breaches, including remediation and reputational damage, they are better positioned to help assess the actual value of cybersecurity investments.

3. Prioritize cybersecurity in mergers and acquisitions

One area cybersecurity becomes especially critical is in mergers and acquisitions. Assessing a target company’s security posture is a critical component of due diligence, and can help create a roadmap for integrating the target company into the acquirer’s security and compliance posture.

This approach includes non-negotiables for day one, such as issuing new, compliant laptops, planning network segregation, and a remediation roadmap for any existing vulnerabilities. Third-party assessments also have a role to play here to help inform post-acquisition plans.

4. Create a cyber-aware culture from the top down

We’ve been vocal about how creating a cyber-aware culture starts at the top. Boards should set the tone by regularly placing cybersecurity on the agenda at the main board level at least once a year.

They can also review internal and third-party attestations, and examine breach action plans to encourage a holistic approach to cybersecurity. Executive leadership must champion the security-first mindset, setting clear expectations, allocating necessary resources, and holding teams accountable. This top-down approach sends a powerful message that security is a non-negotiable priority.

Why boards should have more AI cyber-awareness

Cybersecurity has emerged as a board-level issue because of digital transformation and the emergence of AI, and this presents an opportunity and a challenge. By becoming bilingual in AI and security, boards can ensure their companies are moving decisively to not only improve efficiency and security, but to redefine what’s possible in their industries.

For more on Google Cloud’s cybersecurity guidance for boards of directors, you can check out the resources at our insights hub.

In case you missed it

Here are the latest updates, products, services, and resources from our security teams so far this month:

• Blocking shadow agents won’t work. Here’s a more secure way forward: Shadow IT. Shadow AI. It’s human nature to use technology in the most expedient way possible, but shadow agents pose great risks. Here’s how to secure them, and your business. Read more.
• How to combat bucket-squatting in five steps: Threat actors target cloud storage buckets to intercept your data and impersonate your business. Here’s five steps you can take to make them more secure. Read more.
• How to secure your remote MCP server on Google Cloud: Here are five key MCP deployment risks you should be aware of, and how using a centralized proxy architecture on Google Cloud can help mitigate them. Read more.
• The global harms of restrictive cloud licensing, one year later: Microsoft’s restrictive cloud licensing has harmed the global economy, but ending it could help supercharge Europe’s economic engine. Read more.
• Introducing DNS Armor to mitigate domain name system risks: Google Cloud is partnering with Infoblox to deliver Google Cloud DNS Armor, a cloud-native DNS security service available now in preview. Read more.
• Solve security operations challenges with expertise and speed: At Google Cloud, we understand the value that MSSPs can bring, so we’ve built a robust ecosystem of MSSP partners, specifically empowered to help you modernize security operations and achieve better security outcomes, faster. Read more.
• New GCE and GKE dashboards strengthen security posture: We’ve introduced new, integrated security dashboards in GCE and GKE consoles, powered by Security Command Center, to provide critical insights. Read more.

Please visit the Google Cloud blog for more security stories published this month.

Threat Intelligence news

• Backdoor BRICKSTORM enabling espionage into tech and legal sectors: Google Threat Intelligence Group (GTIG) is tracking BRICKSTORM malware activity, which is being used to maintain persistent access to victim organizations in the U.S. across a range of industry verticals, including legal services, software as a service (SaaS) providers, business process outsourcers (BPOs), and technology companies. The value of these targets extends beyond typical espionage missions, potentially providing data to feed development of zero-days and establishing pivot points for broader access to downstream victims. Read more.
• Widespread data theft targets Salesforce instances via Salesloft Drift: An investigation into Salesloft Drift has led Google Threat Intelligence Group (GTIG) to issue an advisory to alert organizations about widespread data theft from Salesloft Drift customer integrations, affecting Salesforce and others. The campaign is carried out by the actor tracked as UNC6395. We are advising Salesloft Drift customers to treat all authentication tokens stored in or connected to the Drift platform as potentially compromised. Read more.

Please visit the Google Cloud blog for more threat intelligence stories published this month.

Now hear this: Podcasts from Google Cloud

• The AI future of SOAPA: Jon Oltsik, who coined Security Operations and Analytics Platform Architecture (SOAPA), gives hosts Anton Chuvakin and Tim Peacock an update on the ongoing debate between consolidating security around a single platform versus a more disaggregated, best-of-breed approach — including how agentic AI has changed the conversation. Listen here.
• The AI-fueled arms race for email security: Email security is a settled matter, right? Not if AI has anything to say about it. AegisAI CEO Cy Khormaee and CTO Ryan Luo chat with Anton and Tim on how AI has upended email security best practices. Listen here.
• Cyber Savvy Boardroom: Enterprise cyber leadership: Francis deSouza, chief operating officer, Google Cloud, joins Office of the CISO’s Nick Godfrey and David Homovich to talk about the biggest challenge facing boards in the next three to five years: governing agentic AI. Listen here.
• Defender’s Advantage: How vSphere became a target for adversaries: Mandiant Consulting’s Stuart Carrera joins host Luke McNamara to discuss how threat actors are increasingly targeting the VMware vSphere estate, and leveraging in this environment to conduct extortion and data theft. Listen here.
• Behind the Binary: Inside the FLARE-On reverse-engineering gauntlet: Host Josh Stroschein is joined by FLARE-On challenge host and author Nick Harbour, and regular challenge author Blas Kojusner, for an in-depth tour of its history, and discuss how it has grown into a must-do event for malware analysts and reverse engineers. Listen here.

To have our Cloud CISO Perspectives post delivered twice a month to your inbox, sign up for our newsletter. We’ll be back in a few weeks with more security-related updates from Google Cloud.