Google I/O: Decoding Gemini app, AI in Search, Google Beam and Workspace updates

Google’s two new AI subscription plans shouldn’t be a surprise since there is pressure on the widening AI tools to generate revenue

Mountain View, California: With a slew of updates for Gemini 2.5 Pro and Gemini 2.5 Flash, significantly improved generative AI models Veo 3 and Imagen 4, upgrades for Gemini Live, the introduction of Deep Research and Canvas, as well as the Google Al Pro and Google Al Ultra plans going Live, it does lead us to an important question — how is the Gemini app changing? The answer is quite simple.

The first glimpse of Google Beam, and most may not have realised it then, was when Google demoed Project Starline at an I/O keynote a few years ago

The change is rather significant.

First things first, Google has confirmed that the Gemini Live capabilities were now available on all compatible Android and Apple devices, to everyone and without any subscription plan. The arsenal of new tools that have been added to the Gemini app includes the Imagen 4 image generation model, the Veo 3 video generation model (both of these will be in the drop-down list for model selection), the new Deep Research and Canvas features, as well as Gemini finding an integration within the Chrome web browser.

The Gemini 2.5 Flash model now becomes the default model, succeeding the 2.0 Flash model.

“With the Gemini 2.5 models, Canvas is now even more intuitive and powerful. You can create interactive infographics, quizzes and even podcast-style Audio Overviews in 45 languages. But the magic of 2.5 Pro is its ability to translate complex ideas into working code with remarkable speed and precision. People are rapidly bringing entire applications to life from simple descriptions. Vibe coding like this dramatically lowers the barrier to creating software and makes prototyping new ideas faster than ever before,” said Josh Woodward, vice president, Google Labs and Gemini.

Google is introducing two new AI subscription plans, and this shouldn’t be a surprise, since there is pressure on the widening AI tools to generate revenue for the tech giant. There is the Google Al Pro (this is essentially a renamed version of the current Google AI Premium plan, with some add-ons), and Google Al Ultra that would be available as an option for subscribers.

With the Pro plan, users will get a full suite of AI products with higher rate limits compared to their free version, including the Gemini app that was formerly known as Gemini Advanced, alongside products such as Flow and NotebookLM with higher rate limits.

The Ultra plan, as the name suggests, is being touted as the flagship tier, and for now, is available only in the US (some of the functionality in this is limited to the US region, for now). This will have the highest rate limits, early access to new experimental features and priority access to the upcoming Deep Think model as well as the Agent Mode when it is launched. 

“Agent Mode seamlessly combines advanced features like live web browsing, in-depth research and smart integrations with your Google apps, empowering it to manage complex, multi-step tasks from start to finish with minimal oversight from you,” said Woodward.

Google says the Ultra plan costs $249.99 per month, and more countries will be added to the rollout soon. OpenAI also has a Pro subscription that costs $200 per month. Anthropic also has a Max plan for Claude users, which is priced upwards of $100 per month, depending on how it is configured. The India pricing for the Ultra plan remains unannounced. 

AI in search, and the agent aspirations

For Google to morph Gemini into a universal AI assistant, the data that they collect from Search, will be crucial. AI Overviews, which was launched at last year’s I/O, has since seen rollout in more countries such as India. Google said the search queries are on an upward trajectory. AI Overviews in Google Search are now available in 200 countries, and can be overlaid on search results in more than 40 languages.

This year, Search gets an AI Mode. The keys here are advanced reasoning and multimodality. Liz Reid, who is vice president, Head of Google Search, explained that AI Mode will use the query fan-out technique, to break down any question asked by a user, into further subtopics.

“This enables Search to dive deeper into the web than a traditional search on Google, helping you discover even more of what the web has to offer and find incredible, hyper-relevant content that matches your question,” said Reid.

There will also be a Deep Search in AI Mode, which uses the same query fan-out technique. In AI Mode, Google said Deep Search can further issue hundreds of searches, reason across disparate pieces of information, and create an expert-level fully-cited report in just minutes.

Joining visual search pursuits alongside Google Lens is Search Live, which will allow a user to point the phone’s camera at anything around them to begin a search. “For example, if you’re feeling stumped on a project and need some help, simply tap the “Live” icon in AI Mode or in Lens, point your camera, and ask your question. Just like that, Search becomes a learning partner that can see what you see — explaining tricky concepts and offering suggestions along the way, as well as links to different resources that you can explore — like websites, videos, forums and more,” said Reid.

Agentic AI capabilities are getting a profound runaround in the AI Mode, which Google said, can help people save time with tasks such as keeping tabs on and purchasing movie tickets. “This will start with event tickets, restaurant reservations and local appointments. And we’ll be working with companies like Ticketmaster, StubHub, Resy and Vagaro to create a seamless and helpful experience,” the company said. This should grow rapidly for Google, in due course.

As should the AI Mode shopping experience, which uses Gemini as the underlying model, for the Shopping Graph to help users browse for inspiration, think through considerations, and narrow down products to a more manageable shortlist.

“The Shopping Graph now has more than 50 billion product listings, from global retailers to local mom and pop shops, each with details like reviews, prices, color options and availability. And you know you’re getting fresh and accurate information you can trust, because every hour more than 2 billion of those product listings are refreshed on Google. Say you tell AI Mode you’re  looking for a cute travel bag. It understands that you’re looking for visual inspiration and so it will show you a beautiful, browsable panel of images and product listings personalised to your tastes,” explained Lilian Rincon, vice president, Consumer Shopping Product. 

This AI agent uses Google Pay to complete the order if the pricing and other criteria match the checklist you’d set initially.

Beam it, in 3D

The first glimpse of Google Beam, and most may have not realised it then, was when Google demoed Project Starline at an I/O keynote a few years ago. The 3D communication platform, as Google Beam is being called, uses an AI volumetric video model which makes these calls appear fully 3D from any perspective. It transforms standard 2D video streams into realistic 3D experiences. This can transform otherwise 2D video calls into something potentially more immersive, without needing to wear any 3D glasses or virtual reality headsets.

“We're working in collaboration with HP to bring the first Google Beam devices to market with select customers later this year. In just a few weeks, you’ll see the first Google Beam products from HP at InfoComm. We’re also working with industry leaders like Zoom and key channel partners such as Diversified and AVI-SPL to bring Google Beam to businesses and organisations worldwide,” said Andrew Nartker, general manager, Google Beam.

AI in your workspace

Google is not slowing down on wider AI-driven functionality integration within Workspace. They say Workspace delivers 2 billion AI assists every month. Some of the key changes now include the availability of Imagen 4 for generating an image in Slides, Vids and Docs, source-based writing by pointing Google Docs and Gemini to multiple documents where information sources may be scattered, transforming presentation slides into videos, speech translation in Google Meet and inbox cleanup as well as fast appointment scheduling integrated into Gmail.