👋 Hello hello,
Kushank was at the Google I/O event this week. And we have a lot to share.
Google didn't drip-feed features or save the big announcements for a quiet blog post. They put everything on the stage at once.
In a single keynote:
A model that turns a photo or audio clip into an entirely new video scene.
A personal AI agent that keeps executing tasks after you close your laptop.
And what Sundar Pichai called the biggest redesign of Search in 25 years.
All running on a new, faster Gemini.
If last year felt like Google steadying itself, this week looked like a team that's been building in the dark for a while and finally turned on the lights. Let's get into what actually landed.
🔥🔥🔥 Three Updates from Google I/O
Gemini Omni is Google DeepMind's new model built around one idea: take any input, produce any output. The starting point is video, and what it can do is genuinely different from what's been available before.
Give it a photo, a video clip, or an audio file and it builds entirely new scenes from them. It combines Gemini's reasoning with Google's generative media systems, so it actually understands the content — not just the surface level. You can hand it your own footage and iterate on ideas without touching an edit timeline.
The practical implication: this is closer to a creative production tool than a glorified filter.
Gemini 3.5 Flash is Google's new model built for efficiency without sacrificing capability. Fast enough for everyday tasks, smart enough for complex multi-step work, and capable of navigating real-world complexity without slowing down.
It's also the engine powering both Spark and the new Search features, which means it's quickly becoming Google's core workhorse for anything that needs to move fast. Available now in the Gemini app.
Gemini Spark is Google's new 24/7 personal AI agent. Built on Gemini 3.5 and running on Antigravity — Google's infrastructure layer for long-running background tasks — it handles multi-step work without you needing to stay in the loop at every stage.
The detail that separates it from existing AI assistants: Spark runs on dedicated virtual machines in Google Cloud. Your laptop can be closed, your phone on silent, and it's still working through the task list. It connects to Google's tools natively, with third-party MCP integrations on the way.
Think of it as giving your AI its own computer to work from — completely separate from yours.
🧪 PAI Labs
Set it up once. Watch it publish for months.
If you create content, here's what this workflow does: you drop an idea into Notion, and the system takes it from there. Research, drafting, scheduling, formatting — handled. No more staring at a blank doc at 9pm. No more "I'll write it tomorrow." No more bottlenecks.
The whole thing runs on Notion and AI, and once it's set up, it genuinely does the heavy lifting for you. One afternoon of setup, and your content pipeline basically runs on autopilot.
🔥🔥 Two New Google Tool Features
1. 🎥 Ask YouTube
Ask YouTube brings a conversational search experience to YouTube's library. Type a question, and instead of a wall of thumbnails to scroll through, you get relevant videos alongside structured text responses in one place — an actual answer, not a guessing game.
For anyone who uses YouTube as a research or learning tool, this changes how useful it actually is. Rolling out this summer in the US first.
Three things are changing inside Search. A new AI-powered search box that understands context and intent at a much deeper level. Background information agents that surface relevant results 24/7 without you needing to prompt them. And agentic coding capabilities that let Search build custom interactive experiences — dashboards, trackers, simulations — directly from what you typed.
That last one is the real shift. Search has been a lookup tool for 25 years. Now it builds things. Sundar Pichai called it the biggest upgrade to Search in a quarter century. Hard to argue with that framing based on what was shown.
🔥 Things You Didn’t Know You Could do With AI
One of the most underreported announcements from Google I/O this week: Search can now generate custom interactive visuals directly from your question. Powered by Gemini 3.5 Flash and Google Antigravity, Search reads your query, designs the layout, builds the relevant components, and deploys a live interactive result on the spot — tailored to exactly what you asked.
Here's how to use it when it rolls out to your region:
Open Google Search and type a question that benefits from a visual — something like "show me how compound interest grows over 30 years" or "compare fuel efficiency across 2025 SUV models."
Let Search generate the custom interactive output. It handles the layout and builds the components based on your specific query.
Interact with the result directly inside the page — adjust sliders, toggle variables, or explore the data without opening another tab.
If it needs refining, rephrase the question with more specifics. Search rebuilds the visual to match.
Bookmark or share the result link if you want to return to it later.
💬 Quick poll: What's an AI workflow you tried that ended up being more hassle than it was worth?
Don't forget to rate today's post
Until next time,
Team @PracticalyAI
