10 of the best products and features we love from Google I/O 2025
At this year’s I/O, Google made one thing very clear: AI isn’t just a layer on top of its products anymore. It’s the engine powering everything.
Most of the tech world was on standby this week as Google hosted one of its most anticipated events of the year, Google I/O 2025. Held across Tuesday and Wednesday, the event wasn’t just about flashy demos or product teases, it marked an obvious turning point in how Google wants us to work, build, and interact with the digital world.
At the centre of it all was Artificial Intelligence, not as an add-on, but as the operating layer of everything Google’s building. From tools that help you virtually try on clothes, to upgrades to Gemini, and tools that help with simple daily tasks like purchasing tickets, this year’s announcements weren’t about far-off futures, they were about stitching AI into the tools millions already use every day. Here are 10 of the biggest announcements from the events.
/1. Shopping gets personal with ’Try It On’
Imagine browsing online for clothes and actually seeing how they’d look on you—not a model, not a mannequin, but your own reflection. That’s what Google’s bringing to Search Labs with its new virtual try-on experiment.
Rolling out to users in the U.S., this feature lets you upload a photo and try on items from billions of apparel listings in real-time. It’s powered by generative AI and style-matching algorithms, helping you see fit, colour, and vibe before you ever hit “buy.” If online shopping ever felt like guesswork, this is Google handing you a virtual changing room.
/2. New Canvas tool to help transform your texts
But Google’s creativity tools aren’t just dressing you, they’re now dressing up your ideas too. The new “Create” menu in Canvas, also launching today, is like a magic box for content makers.
Type in some text, and you can turn it into everything from interactive infographics to sleek web pages, immersive quizzes, or even audio explainers in 45 languages. Whether you’re a teacher building lesson plans or a marketer sketching out a campaign, Canvas wants to do the heavy lifting—no design skills required. It’s part productivity booster, part creativity sidekick.
/3. Gemini just became your study buddy
And speaking of quizzes, Gemini is getting smarter, and a little more fun. You can now ask it to “create a practice quiz on…” pretty much any topic, and it’ll generate interactive questions instantly.
For students, this could be a faster revision session and customised testing without endlessly Googling flashcards. For parents, it may be an easy way to keep kids engaged with meaningful learning. For everyone else, it could be a glimpse at how generative AI can actually teach, not just tell.
/4. Meet Flow, the filmmaker in your browser
The creativity upgrade doesn’t stop with text and quizzes, Google’s rolling out Flow, an AI-powered filmmaking tool built with DeepMind’s top-tier models. It’s not your average video editor. Flow lets you storyboard, style, and direct cinematic clips with AI handling the heavy lifting.
You can guide how characters move, what scenes look like, even how the camera shifts—making it accessible for creators without a Hollywood budget. If Veo brought the visuals, Flow adds control and narrative. It’s filmmaking democratised.
/5. Music AI Sandbox: now in more hands
And while visuals got a boost, so did sound. Google is expanding access to Music AI Sandbox, powered by its upgraded music model Lyria 2. Originally only for select creators, it’s now available more widely through YouTube Shorts and to enterprise users via Vertex AI.
This tool allows users to compose, experiment, and remix with AI-generated music that feels anything but generic. Whether you’re building a jingle, background score, or full-blown track, Sandbox gives you the ingredients and the freedom to explore.
/6. Agent Mode turns Gemini into your personal assistant
After months of speculation, Google finally unveiled Agent Mode for Gemini: a new way to get things done by simply stating your goal. Instead of prompting for information, users can ask Gemini to take actions, like planning trips, managing tasks, or summarizing files, without micromanaging every step. It’s not just chat anymore; it’s delegation.
While it’s still experimental and limited to Google AI Ultra subscribers, Agent Mode signals a shift in Gemini’s evolution from passive assistant to autonomous operator. That move mirrors what Microsoft is attempting with its Copilot agents, but Google’s bet here is stronger integration with Android, Search, and your Google apps.
/7. Astra shows off real-time AI that sees, speaks, and remembers
Project Astra was one of the flashiest demos of I/O. It’s Google’s prototype for a universal AI agent that can interact in real time with the world through your camera and voice. Think: pointing your phone at a bike lock and asking how it works, or having a back-and-forth conversation while the agent remembers your context.
This kind of real-time, multimodal memory is where AI assistants could become truly useful, and it puts Google on track to compete directly with OpenAI’s GPT-4o and Microsoft’s Teams-based agents. But unlike Microsoft’s workplace-first approach, Astra seems more aimed at daily life.
/8. Gemini 2.5 Pro gets a brain boost with Deep Think and security upgrades
Google’s flagship model got two big updates. First, Deep Think—an opt-in feature that allows the model to take extra time to reason through complex queries. Second, enhanced security tools to guard against prompt injection and malicious use.
Together, these upgrades position Gemini 2.5 Pro as a serious enterprise option and reinforce Google’s push for safer, more capable AI. While OpenAI and Anthropic have touted reasoning, Google bets that letting users control the “thinking budget” will give them more trust and flexibility.
/9. Gemini Live integrates with your Google apps
Gemini Live, Google’s voice-first assistant, is now getting smarter by connecting directly to apps like Calendar, Maps, and Keep. This means you’ll be able to ask it to reschedule meetings, pull up directions, or save notes in real time, during your voice interactions.
It’s a small feature on the surface, but this integration is crucial. By anchoring Gemini inside your workflow, especially on mobile, Google is quietly making its AI indispensable. And for users comparing Gemini to ChatGPT or Copilot, app-level actions could be a major deciding factor.
/10. Project Mariner brings agentic AI into Search
Last but not least: Google Search is becoming agentic. With Project Mariner, Search’s new “AI Mode” can now help book reservations, buy event tickets, or schedule local appointments directly from your queries. It’s the kind of hands-on execution that turns Search from a lookup tool into a utility layer.
If Microsoft made Office the place where AI lives, Google is making Search the launchpad—and that has massive implications for how billions of people interact with AI every day.
Conclusion
Google I/O 2025 made one thing clear: the future of AI won’t be just about clever replies or image generation. It’s about building agents that remember, act, and integrate deeply into your life. Google’s announcements feel like the groundwork for a more personal, more useful AI experience, and the race to own the agentic web is officially on.
Oluwaseun Bamisile
