If there was any doubt left that Google is now a full-bore AI company, I/O 2025 just put it to rest.
This year’s keynote was a clean break from the past. Gone were the typical updates on Android, Pixel, or Maps. Instead, every second of the two-hour presentation was laser-focused on Google's evolving AI arsenal: Gemini, Astra, Flow, Imagen, Veo, and more. In short, this wasn't a tech conference. It was an AI coming-out party.
Here are the biggest takeaways from a future-forward showcase that might just redefine how we interact with information, media, and the web itself.
1. Search is No Longer a Box. It's an AI Interface
In the post-ChatGPT internet, "Googling" isn't the reflex it used to be. Google knows it and it’s adapting fast.
AI Mode for Search is now officially rolling out to all users in the US. It adds a conversational layer to the search experience, complete with new tools in Search Labs like:
-
Deep Search for long-form queries and research projects
-
Project Mariner, a real-time ticket-buying assistant
-
Wider rollout of the (controversial) AI Overviews, now more embedded than ever
This isn’t just window dressing. It’s a structural change in how Google wants us to access and act on information, faster, more visually, and with less typing.
2. Shopping Gets Personal and a Little Bit Sci-Fi
Google’s AI Mode now responds to open-ended shopping prompts like "Find me a cute purse" with curated visual suggestions. That alone would be a UX win, but it gets better.
Enter virtual try-on, a Gemini-powered feature that lets users upload a single image of themselves and see how clothing items would look on their body. It's still in Search Labs for now, but the potential is clear: a shopping experience built around you, not around filters.
And for checkout? Google's new agentic buying feature can auto-purchase your item when it hits your preferred price. It's like Honey and a personal assistant had a baby.
3. Gemini Everywhere, from Browsers to XR Glasses
Google’s Gemini AI has officially permeated the ecosystem.
-
Gemini in Chrome can summarize web pages, explain technical info, and eventually navigate across tabs and sites for you
-
Gemini XR, coming to headsets and smart glasses, brings live translation, object recognition, and location-based assistance
-
Gemini Canvas, now more intuitive, can turn prompts into code, infographics, games, and even full websites without needing dev skills
And for power users? Google launched Gemini AI Ultra, a $250 per month subscription with premium tools like Deep Think Mode and high-res generation in Imagen 4. It’s a move that clearly courts creatives, devs, and pros at the bleeding edge.
4. Veo and Flow: AI Video Enters the Talkies Era
AI-generated video is no longer a proof of concept. It’s a toolset.
Google’s new Veo 3 model doesn’t just create video clips. It adds synchronized sound, dialogue, and atmosphere. Think Sora or Pika, but with the audio baked in.
And with Flow, Google introduced a generative filmmaking suite where you type in a scene and get a full cinematic asset: characters, environments, the works. Want to see surgeons in the back of a 1970s taxi? Done.
This could reshape pre-production workflows or even democratise filmmaking entirely. Whether you’re mocking up a concept or scripting a pitch, AI video just got dramatically more usable.
5. Beam and Astra: A New Kind of Presence
Forget Zoom fatigue. Google’s Beam is turning video calls into 3D experiences with no VR headset required. It builds on the long-rumoured Project Starline, using cameras and AI to project you into a virtual shared space. The caveat is that it still needs specialised booths and will roll out to early enterprise customers first.
Meanwhile, Project Astra — arguably the most sci-fi announcement of the lot — is becoming your real-time visual co-pilot. Need to fix a bike? Point your camera. Want to know what flower that is? Just ask. Astra combines vision, reasoning, and natural language in ways that feel genuinely next-gen.
Final Thought: Google Isn’t Just Competing with OpenAI. It’s Rewriting the Interface Layer
What Google I/O 2025 made clear is that this isn't about beating ChatGPT feature-for-feature. It’s about owning the new default way people get things done — shopping, researching, building, watching, creating.
With Gemini baked into nearly every layer of its ecosystem, Google is repositioning itself not just as an AI player, but as the AI operating system for everyday life.
Whether users opt in or even realise they’re opting in is a different story. But either way, the era of “search, type, click” is drawing to a close.
And Google, it seems, is already writing the manual for what comes next