I was able to attend day 1 of Google I/O today and there were a lot of really cool announcements. Here's my run-down. I took special note of what's actually available to try, and what's coming soon. I'll share deeper product thoughts in the weeks ahead.

This year was about 95% about AI and 5% about Android XR. I felt like there was more optimism and creativity than there has been in the past couple years, which was a nice return to form for Google IO.

Main Keynote

  • Google starlight, Googles 3d vision-enabled video chat is now Google beam. It offers 3d displays 60fps smooth experiences in realtime, allegedly. I haven't seen or been able to try it out.

    • When? Coming soon. For sale this year via HP.

  • Google translate in meet.

    • When? Coming soon. More languages.

  • Project Astra now in Gemini live.

    • When? Announced as available now, but really more of a "experiments as they graduate situation"

One of my favorite moments of the event was when they had a snarky interaction with AI. A woman kept asking silly questions or insisting things were happening that weren't. She claimed some palm trees were small, that her shadow was following her, and that an electrical box was a package delivered to her lawn.

Instead of getting confused or hallucinating, the highly-prepared canned video demo gave somewhat snarky direct responses and rejections, which was a fun comedic moment.

  • Project mariner - interact with web. Agents.

    • When? Now multitasking (10 task)s
    • Teach and repeat learning
    • Coming to Gemini api.
    • Coming soon this summer

  • Agent to agent protocol

  • Gemini sdk compatible with MCP tools

  • Gemini 2.5 flash new version coming soon June.

    • When? Try now in preview.

  • Native audio output (expressive) - more natural and expressive spoken language generation. Wasn't demo'd a ton.

    • When? Available in Gemini api today

  • Thinking budgets - a breakdown of how the model is thinking about a problem

    • When? coming soon in coming weeks

I thought it was really funny when they were building a 3d visualization of art and they talked about "2d web design" as if that was legacy or a bad way of doing or thinking about things.

  • Jules.google (yet another vibe coding / editor)

    • When? now available in public beta

  • Gemini diffusion

    • Text diffusion model - 5x faster than Gemini 2.5 Flash Lite
    • When? Coming soon

  • Gemini 2.5 pro deep think

    • Slower better results
    • Blocked on frontier safety evaluations
    • When? Trusted testers via Gemini API available now

  • World model - simulate aspects of the world

    • Virtual worlds and theme parks
    • Genie 2 interactive games from prompt
    • Veo understands gravity, light, physics (allegedly)

  • Gemini robots in AI sandbox

  • Google's vision for the Gemini app is to be a Universal AI assistant

  • Google search AI mode - longer queries, follow up

    • When? Available now for everyone in the US
    • Personal context in AI mode coming soon this summer
    • When? Deep search coming soon this summer
    • When? Complex analysis and data vision coming soon this summer
    • When? Project mariner agentic AI coming soon this summer

  • Lens now has 1.5bn monthly active users.

  • Project Astra coming to AI mode.

    • Search live. Back and forth in realtime
    • Like a video call with search
    • When? Search live coming soon this summer

  • Virtual try on

    • When? Visual shopping coming soon in the months
    • When? Try on virtually now available today

  • Gemini live now includes camera and screen sharing

    • When? Available now
    • When? Coming soon connected to other apps like drive and keep

  • Project Astra features graduate into Gemini live

    • When? Deep research starting today.
    • When? Upload your own files. Coming soon
    • When? Coming soon search drive and Gmail

  • Canvas - interactive space for code creation.

    • When? Introducing Gemini in chrome coming soon this week in the US to Gemini subscribers
    • When? Available today latest image generation imagen 4

I'm definitely going to try our Lyria to score a vlog at some point.

  • Lyria 2 for YouTube creators vertex API.

    • When? Available now.

  • SynthID. watermark. GetReal and Nvidia detection Available now

  • Veo + imagen for creative filmmaking = flow

    • When? Available now

  • Google AI pro 19.99 per month

  • Google AI ultra 249.99 per month. 50% off for 3 months

  • Android XR

    • Someday they want Gemini on watch car tv, Android XR.

  • Google Glass is back baby in the form of glasses with a heads up display. I have lots of thoughts!

    • Gentle monster and warby parker building glasses with XR
    • When? Coming later this year

  • Project Moohan - First Android XR.

    • When? Available later this year. Coming soon.

  • Agents in chrome, search, Gemini.

  • Agent mode - planning and scheduling tours.

  • Keep searching for new listings

  • Agent mode coming soon

  • Personal context

    • Responses that sound more like you .
    • Tone style docs.
    • When? Coming soon to Gmail this summer

  • Veo 3

    • When? available now
    • Includes audio

Developer Keynote

The developer keynote attempted to be more honest, accurate, and achievable. They stated a goal of spending the first hour doing demos, but then they ended up talking to slides and videos for at least half of that time.

Many of the demos were also not real coding demos. They loaded some pre-baked code and then hit the big AI button and hoped for the best.

  • Google stitch - Another AI editor / vibe coding, with designer
  • Jetpack compose
    • camerax beta
    • Media APIs in GA
  • Android AI studio tool coming soon to help dependency updates.

There was so much more but I wanted to start sharing this right away. Let me know if any of this was helpful!