Revolgy blog

Our top takeaways from Google Cloud Next 25

Written by Jana Brnakova | April 16, 2025

Spoiler: AI was everywhere

Google Cloud Next wrapped up its 2025 event in Las Vegas, and there’s still a lot of excitement about the new things announced. This year’s event looked closely at how cloud technology is changing quickly, with a big focus on how AI can change things. For us at Revolgy, Next 25 was a great chance to see the newest advancements that will shape how businesses work and grow in the cloud in the future.

Speaking of growth and working together, we’re very happy to announce that Revolgy was named the 2025 Google Cloud Country Partner of the Year for Central and Eastern Europe (CEE). This is our fifth award from Google Cloud, and it recognizes how hard we work to help businesses in the CEE region use the cloud, get the most out of it, and succeed in the long run.

Revolgy’s CEO, Miroslav Vlasák, and Chief Architect, Josef Bárta, receiving the Partner of the Year Award

 

So, what were some of the most interesting announcements from Google Cloud Next 25 that we’re particularly enthusiastic about? Here are a few key points that combine Google’s new ideas with the potential for businesses in the cloud to change, including how Revolgy can help you use them.

Watch the Google Cloud Next 25 opening keynote in 10 minutes.

Gemini shows off strong new models

There was a lot of talk about Gemini, Google’s most advanced AI model family. We saw the announcement of Gemini 2.5 Pro, now available for public testing, designed for handling tough tasks that need deep thinking and coding skills. It’s even ranked #1 on Chatbot Arena.

Alongside it, Google announced that Gemini 2.5 Flash, a model that responds quickly and is cost-effective, is coming soon. This means AI is getting smarter and more useful, allowing businesses to do things like analyze data in more detail and respond to customers faster As a partner with a lot of experience in AI, Revolgy is ready to help businesses use these powerful Gemini models in their workflows.

Agentspace brings AI to daily work

To make sure AI tools can benefit everyone — not just developers — Google introduced some big updates to  Agentspace. This platform is built to help employees across an organization work faster and more efficiently by adding AI right into the tools they already use, like Chrome Enterprise.

 

 

With Agentspace, people can quickly find information, documents, and internal resources without needing to ask IT or dig through different systems. One of the most exciting features is the new no-code Agent Designer, which allows users to create custom agents that automate everyday work tasks.

Agentspace is a great starting point for companies that want to make AI more accessible to everyone on the team, not just engineers.

At Revolgy, we can help businesses set it up to work with their existing tools and support their day-to-day operations, solving real business problems.

Smarter cybersecurity

Google introduced Google Unified Security, a new solution that brings together several advanced security tools into one platform. It combines things like threat detection, real-time alerts, and AI assistance, built on a global data foundation designed to handle large volumes of information.

One standout feature is the alert triage agent, an AI assistant that automatically investigates security alerts to help teams respond faster and reduce routine work. For businesses looking to strengthen their cybersecurity using smart, modern tools, Revolgy can help set up and manage Google Unified Security in a way that fits your environment.

Updates to Vertex AI

Google Cloud’s complete AI platform, Vertex AI, received some major updates. Businesses can now more easily train and fine-tune powerful models like Gemini and Llama using their own data, so the AI actually learns from what’s most relevant to them, not just public datasets.

A new feature called Vertex AI Dashboards makes it easier to track how models are performing — how much you’re using them, how fast they respond, and how reliably they run.

One of the most forward-looking additions is support for multi-agent systems. Instead of relying on a single AI model to do everything, multi-agent systems allow multiple specialized agents to work together on more complex tasks, like a team. Google introduced the Agent Development Kit (ADK) to help developers create these networks of agents faster. There’s also a new open protocol called Agent2Agent (A2A), designed to let agents built on different platforms or by different vendors communicate securely and effectively.

Video, image, speech, and music

Besides large language models, Google is also building new tools for generative media, giving creative teams and businesses new ways to improve how they produce content.

Imagen 3: Enhanced text-to-image generation

 

Imagen 3 is Google’s best text-to-image model yet, designed to create highly realistic and detailed images from written prompts. It does a great job of understanding and following even complex instructions, which means your ideas are translated into visuals with impressive accuracy.

Chirp 3: Fast, custom voice and smarter transcriptions

Google’s advanced model for audio and speech-related tasks, its most exciting feature is Instant Custom Voice, which lets you create a custom, natural-sounding voice using just 10 seconds of audio. 

Chirp 3 also improves how audio is transcribed, especially with multi-speaker recordings. It can now better tell who’s talking and separate overlapping voices, which is a big help for meetings, interviews, or support calls.

Lyria: Generative AI for music and soundtracks

Lyria is a music generation model, built to turn text into high-quality, structured music. It’s the first enterprise-ready text-to-music model available on Vertex AI, and it allows users to generate coherent melodies, harmonies, and full musical compositions from simple descriptions.

It also supports instrument separation, meaning different parts (like vocals or drums) can be isolated and edited more easily — making it a powerful tool for music production and custom sound design.

Veo 2: New set of editing features

Google’s upgraded generative video model, Veo 2 is made for creating high-quality videos from text prompts. It produces realistic, high-resolution footage that reflects the fine details in your descriptions — from environment and actions to camera angles and motion.

 

Clean, professional edits without manual retouching

 

One standout improvement is camera control, allowing creators to simulate pans, zooms, and other cinematic movements directly from a prompt.

 

Interpolation creates smooth transitions across frames

 

Veo 2 also includes video editing capabilities, so you can fine-tune or revise AI-generated content without starting from scratch.

New AI chip hits hyperspeed

To keep up with the rising demand for AI computing power, Google introduced Ironwood, its most powerful custom-built AI chip yet. It’s part of their seventh-generation Tensor Processing Units (TPUs) — special hardware designed to run large AI models faster and more efficiently.

Compared to the first version Google made available to the public, Ironwood is an incredible 3,600 times faster. Each setup, or “pod,” contains over 9,000 of these chips, delivering massive computing power to handle today’s most complex AI tasks.

Smarter coding with Gemini Code Assist

Developers now have a new AI assistant to help them code more efficiently: Gemini Code Assist. The tool can assist with everyday coding tasks like migrating code, adding new features, reviewing code changes, and generating tests.

It also includes a built-in Kanban board, making it easy to track the progress of AI-assisted tasks. It’s now available in Android Studio and works with other popular developer tools from companies like Atlassian, Sentry, and Snyk.

Bringing Gemini to your location with Google Distributed Cloud

For organizations with strict rules or data location needs, Google has made it so that Gemini can now run on their own computers and data centers, not just on Google’s cloud. This brings Google’s advanced AI models to on-premises environments, and it’s all made possible by Google Distributed Cloud (GDC) and partnerships with NVIDIA and Dell.

A growing collection of AI models

The Vertex AI Model Garden continues to grow, now featuring over 200 models, including Google’s own models, models from other companies like Meta and Anthropic, and open models like Llama.

Google Cloud Next 25 highlighted just how far cloud technology, especially AI, is moving.

At Revolgy, we’re excited about what’s ahead. Winning the 2025 Google Cloud Country Partner of the Year for CEE means a lot to us, and we’re ready to help our customers use these new tools, like Gemini and Vertex AI. 

If your team is exploring what’s next in the cloud, we’re here to support you on that journey, with tools that make sense and have an impact. Just let us know!