Google I/O 2025: AI Mode on Google Search, Veo 3, Imagen 4, Flow, Gemini Live, and More

Google’s annual occasion I/O has returned this yr, pushing the boundaries of AI additional than ever earlier than. What began off with the keynote speech by Google CEO Sundar Pichai, highlighting the milestones achieved by the tech big, quickly escalated to an ever-exciting present of AI-powered developments and new generative AI instruments. From the brand new AI mode on Google Search and Gemini Reside to the launch of Veo 3, Imagen 4, and Circulate, to the disclosing of Android XR and Samsung Moohan – Google was pulling one AI rabbit after the opposite out of the hat. Of all that was mentioned and proven, this weblog brings you the 8 greatest AI breakthroughs and launches introduced on the Google I/O 2025 occasion.

1. Google Beam & Actual-Time Translation in Google Meet

Google has taken video calling to an entire new degree with Google Beam – an evolution of Undertaking Starline that gives immersive 3D video communication. This new expertise captures the views of the speaker from 6 completely different digital camera angles, together with their motion at 60 fps. It then places all of them collectively to generate a 3D model of the speaker, making it really feel just like the individual is true in entrance of you. Aimed toward making digital interactions really feel extra lifelike, Google Beam will quickly be accessible to Google Meet customers within the US after which to different nations.

Complementing this, Google Meet now options real-time speech translation. Powered by AI, this translation function can choose up your dialect, tone, and nuances to offer correct translations in real-time. Initially supporting English and Spanish, Google plans so as to add extra languages quickly, facilitating seamless multilingual conversations throughout video calls. This new function has already been rolled out to US customers and can quickly be launched worldwide. Google Enterprise customers may also get entry to this function in the direction of the top of this yr.

2. AI Mode in Google Search

The most important announcement made at Google I/O 2025 has received to be concerning the new AI Mode in Google Search. Owing to the widespread acceptance of AI overviews on Google Search, they’ve now introduced the facility of AI on to the search bar with the AI Mode. This new function lets customers use AI on to seek for outcomes, simply as they’d on ChatGPT, Gemini, or another AI chatbot.

With an expanded search window, customers can now add extra context and ask a number of questions throughout the identical search question. Google Search breaks consumer queries into a number of smaller queries and classes and runs parallel searches on all of them. With AI-powered reasoning capabilities, it then places collectively all the information and generates a complete and contextual response. This transforms Google Search right into a extra interactive expertise.

Key Options

Google Search’s new AI Mode gives 7 new options:

1. Private Context: Now you can get Google to offer you customized responses by integrating your search historical past and knowledge from different Google apps and instruments like Gmail. This integration lets the AI perceive your model and decisions, to generate smarter responses which are uniquely useful to you.
2. Deep Analysis: This function multiples the online search capabilities of Google to do dozens and even lots of of searches on the identical time, to collect extra data, leading to extra detailed and well-researched responses.
3. A number of Response Codecs: The now AI-powered Google Search dynamically generates one of the best format for every response, based mostly on the question. As an illustration, it might probably intelligently generate interactive lists and graphs for sports activities and monetary queries.
4. Customized Purchasing Strategies: As a substitute of merely itemizing out product pages and procuring hyperlinks, Google Search can now provide you with personalized procuring recommendations based mostly in your style, earlier searches, and buy historical past. When you can add extra context and particulars to your search question, Google additionally recommends factors to think about that can assist you make the proper alternative.
5. Digital Outfit Trials: One other spotlight of the AI Mode is the AI-powered procuring with digital try-ons. Now you can nearly strive on garments earlier than shopping for them, straight on Google Search. Merely choose the outfit, add your picture, and watch as Google magically clothes you up in that outfit proper on the display screen. This function has additionally been rolled out to customers within the US right now.
6. Search Reside: Now you can do dwell video calls to Google Seek for real-time visible help, much like the Gemini Reside function on the chatbot.
7. AI-powered Visible Search: The place Google Lens would earlier discover related photos based mostly on the enter picture, it might probably now give AI overviews of any picture you click on or add. It could actually mainly clarify something that’s in entrance of your eyes, being a digital companion, particularly to the visually impaired.

3. Undertaking Astra: Gemini Reside

The most important replace of the Gemini chatbot introduced on the Google I/O occasion this yr was the Gemini Reside function. An extension of Google’s Undertaking Astra, Gemini Reside is developed to be a common AI assistant. It lets customers have dwell video calls with the AI-powered Gemini chatbot, providing real-time AI help for something and every part. It lets customers interact in interactive digital camera conversations, obtain on-the-go translations, and share screens or digital camera feeds for assist. This function is now accessible in over 45 languages throughout 150+ nations, to each Android and iOS customers.

4. Undertaking Mariner: Agent Mode

At Google I/O 2025, the corporate demoed Agent Mode – a Undertaking Mariner-based AI agent with laptop use options. This superagent is able to doing as much as 10 duties without delay. It could actually make calls, search the online, discover YouTube movies, provide you with recommendations, reply questions, and do much more. It’s also clever sufficient to study the workflow from one activity and apply it to do different duties equally, utilizing a expertise referred to as ‘train and repeat’.

Agent Mode is designed to be private, proactive, and highly effective. It could actually entry your calendar, see upcoming occasions, and set reminders or prep you for the occasion, even earlier than you ask it to. This degree of autonomy and intelligence hasn’t been seen in general-use AI brokers earlier than. It helps in automating quite a lot of on a regular basis duties like scheduling, notetaking, interview prep, and so forth.

Agent Mode in Motion

Google took it even a step additional by integrating it with Google Search’s AI Mode to deliver its customers an agentic search function. With this, customers can run a number of net searches and net search-based duties within the background, which can be executed autonomously by the agent.

For instance you should utilize this function to arrange agentic checkout for e-shopping. When you discover a product that you just want to purchase, you should utilize agentic search to seek out it for you inside your funds. Google Search would maintain monitoring the worth throughout web sites, and as soon as the worth drops to your vary, it can routinely place your order with only one faucet. You possibly can even pay for it utilizing Google Pay, once more with only a faucet.

So now, similar to Agent2Agent Protocol, and Anthropic’s Mannequin Context Protocol, Gemini API and SDK will have the ability to use MCP instruments. Google will quickly roll out Undertaking Mariner’s laptop use capabilities to builders through the Gemini API. In the meantime, an experimental model of Agent Mode will quickly be coming to subscribers on the Gemini app.

Google introduced a few of its newest and most superior generative AI instruments on the Google I/O 2025 occasion. This included:

1. Music AI Sandbox with Lyria 2: The Music AI Sandbox, powered by Lyria 2, allows customers to generate music compositions utilizing AI. It could actually create harmonies, rhythms, background scores, and even full compositions with orchestra based mostly on consumer enter.
2. Genie 2: This superior device from Google can remodel 2D Photographs into interactive 3D environments in simply 2 steps and a immediate. It has a variety of purposes in gaming, digital actuality, and digital content material creation.
3. Imagen 4: Imagen 4 is Google’s newest text-to-image technology mannequin, able to producing high-quality, photorealistic photos from textual descriptions. Not solely does it get textual content and spelling proper, however it might probably additionally intelligently choose the proper font, font measurement, and so forth., based mostly on the question. Furthermore, it really works as much as 10x sooner than earlier fashions.
4. Veo 3: Google launched its newest model of Veo on the annual occasion. The upgraded Veo 3 takes AI-powered video technology to an entire new degree, creating hyper-realistic and high-quality movies from textual content prompts. Together with video, it additionally generates reasonable audio output together with dialogues and background sounds.

5. Circulate: This new filmmaking device from Google brings collectively the inventive capabilities of Veo, Imagen, and Gemini. It permits customers to generate quick movies from textual content or picture prompts, integrating sound, dialogue, and visible results. With text-to-image, image-to-video, and text-to-video options, it turns into a one-stop store for bringing creativeness to actuality. Furthermore, it additionally comes with scene extension and enhancing options.

Most of those updates have been built-in into the Gemini chatbot making it way more superior and creatively succesful.

6. Gemini Integration with Imagen 4, Veo 3, and Extra

Google I/O 2025 was extra about Gemini than about AI, as confirmed by CEO Sundar Pichai’s phrase counter. A number of bulletins relating to Google’s Gemini chatbot have been made on the occasion together with updates on Deep Analysis and Canvas and integrations with Google’s newest generative AI instruments.

Gemini Updates Launched at Google I/O 2025

Right here’s an inventory of all of the Gemini Updates revealed on the Google I/O occasion this yr.

1. Gemini in Chrome: The subsequent large factor is that Google will quickly be rolling out Gemini on Google Chrome as an internet looking AI agent. This lets customers ask their search queries and follow-up questions concerning the search outcomes, on to the AI chatbot.

2. Gemini Voice: Google has built-in native audio output into Gemini’s Voice Mode which lets it reply to customers in a extra customized and nuanced method. It could actually swap between languages, change tones, and even whisper throughout the identical dialog. You possibly can take a look at out this up to date model through the Gemini API.
3. Deep Analysis: Now you can add your individual recordsdata to information the analysis agent whereas doing Deep Analysis utilizing Google Gemini. You too can join it to your Gmail and Google Drive to fetch extra knowledge or present some context.

4. Canvas: The Canvas function on Gemini can now convert deep analysis experiences into customized podcasts, quizzes, infographics, and extra.

5. Imagen 4: Google Gemini’s picture technology capabilities are actually powered by Imagen 4 making the pictures extra reasonable and detailed.
6. Veo 3: Gemini can now generate reasonable movies with correct audio, dialogues, and background sound, because of the newly built-in Veo 3.

7. Android XR and Samsung Moohan

Android XR is Google’s first-ever Android platform, venturing into prolonged actuality. This expertise, powered by Gemini, fosters an immersive expertise for customers via hyper-realistic movies in real-time. Samsung’s Moohan, a newly designed pair of sensible glasses, could be the primary machine that leverages Android XR for AI help. These glasses supply options like real-time navigation, translation, and digital camera live-streaming, aiming to boost consumer interplay with the digital world.

With these glasses you’ll be able to watch dwell occasions from your house, as in case you have been sitting within the entrance row of the stadium. With the flexibility to indicate Google Maps in 3D, it might probably visually take you locations in real-time, providing you with a sensible expertise. Furthermore, it comes with reminiscence and may reply questions. Designed to supply AI help in real-time similar to a human companion, Samsung Moohan can click on photos, make bookings, and even translate audio to textual content. Not like most different sensible glasses that are available in a single sci-fi impressed design, these ones are going to be designed in varied types by Light Monster and Warby Parker.

8. New Google AI Subscription Fashions

Aside from all these launches and updates, Google additionally launched two new subscription plans at its annual occasion:

• AI Professional: Priced at $19.99/month, this plan gives customers all its superior AI options with excessive limits. That is appropriate for basic customers worldwide.
• AI Extremely: Priced at $249/month, this plan provides entry to probably the most superior AI instruments and fashions, together with experimental options, with increased charge limits. That is focused at professionals and enterprises within the US solely.

Conclusion

Google I/O 2025 was fairly the present, giving us all a glimpse into the formidable AI plans Google has. From enhancing on a regular basis instruments like Google Search and Google Meet to growing superior inventive instruments like Circulate and Genie 2, Google’s improvements goal to redefine the boundaries of AI. As these updates and fashions roll out, I’m positive AI can be an integral a part of a typical man’s on a regular basis life. Be it Undertaking Astra, Undertaking Mariner, or Android XR – these developments mark a major step towards a extra intuitive and immersive digital future, powered by AI.