Google I/O 2025: Spotlight on Creative AI
My favorite announcements and moments
Written by Margaret Maynard-Reid, AI/Cloud GDE & 3D Fashion Designer
Google I/O 2025 was full of whirlwind of new announcements as always. This year, it was amazing to see such a lineup of Creative AI announcements at a developer conference.
My top picks are: Gemini for everything, Veo 2 updates, Veo 3 for video generation with native audio, the AI filmmaking tool Flow, Imagen 4 for images, and Lyria 2 for music. These creative AI models / tools offer exciting possibilities for developers, artists, designers, film makers and content creators alike. (Note: learn more about Google for Developers and catch up on all Google IO announcements here. )
In this post, I share my thoughts on the announcements, along with my incredible in-person experience attending Google I/O 2025.
Gemini
Gemini, Gemini, Gemini - has been the highlight throughout IO Keynote. The word ‘Gemini’ was mentioned 147 times and Gemini is everywhere: integrated across the entire Google product ecosystem.
Google Search is evolving with AI, featuring more widespread AI Overviews and the introduction of AI Mode, a reimagined search experience powered by Gemini 2.5.
Gemini Live: Integrates Project Astra’s camera and screen-sharing into the Gemini app for more interactive AI assistance, rolling out on Android and iOS
Veo 3
This is by far my favorite announcement: Google’s new video generation tool Veo 3 creates stunning videos with native audio generation such as singing, music, different sounds and perfect lip syncing.
Since then the internet has been buzzing with all the videos created with Veo 3.
I have been having so much fun creating videos with Veo 3: from Chinese Guzheng music, 1920’s Charleston dance, to the beautiful ballet dance.
You can explore Veo 3 in the Gemini app, Flow and Google Vertex AI.
Veo 2
While the Veo 3 announcement took all the spotlight, don’t forget that Veo 2 new features were made available, for example: camera controls, extend video, keyframe animation, outpainting and object add / remove. In fact, Veo 2 is what’s powering most of the features in Flow (see below).
I have been an early tester of Veo 2 advanced features on Google Cloud Vertex AI.
These features became available in Google AI Studio, Vertex AI and Flow (see below).
Flow
The new AI filmmaking Flow is the evolution of VideoFX, a Google Labs experiment that launched last year. Since August 2024, I have been one of the very first early testers for VideoFX along with other artistis and filmmakers, and I’m super excited to try it out.
It’s powered by both Veo 3 (for video generation with strong prompt adherence and cinematic quality) and Veo 2 for camera controls, Imagen 4 for image generation, and Gemini 2.5 for intuitive prompting.
Flow aims to make it easier for storytellers to explore ideas and create cinematic clips. Users can bring their own assets or generate new ones using Imagen. The tool includes features such as camera controls, Scene builder for editing shots, asset management, and Flow TV for inspiration.
Lyria 2
Unveiled as Google’s advanced AI music generation model, Lyria 2 is capable of creating high-fidelity audio, complex compositions, and rich vocals with fine-grained creative control. This powerful tool is now more accessible through YouTube Shorts and Vertex AI, empowering artists and creators to explore new musical frontiers and overcome creative hurdles.
You can explore Lyria 2 in Google AI Studio or Vertex AI.
Imagen 4
Imagen 4 is Google’s latest image generation model with improved quality and speed, integrated into the Gemini app and also available in Vertex AI.
It has remarkable clarity in fine details like skin, fur and intricate textures, and excels in both photorealistic and abstract styles.
Its text rendering is amazing in terms of spelling and typography. Here is a beautiful Imagen 4 generated vogue magazine cover. The text ‘VOGUE’ accurately reflects the didot font type. I also really like the look: the dress color, the glamour and the pose of the fashion model.
Imagen 4 powers the image creation in the Gemini App and Flow.
Stitch
This year’s Developer Keynote opened with another Google Labs innovation: UI design tool Stitch powered by generative AI! It’s an experimental AI tool that generates UI designs and frontend code for desktop and mobile apps from natural language or image prompts. It aims to accelerate the app creation process and bridge the gap between design and development by allowing for rapid iteration and export to popular design and coding formats.
Read the official Google Labs blog post for more details: From idea to app: Introducing Stitch, a new way to design UIs.
Virtual Try-on
I was very impressed by the Keynote demo of virtual try-on where you can upload your photo and see how the garment looks on you.
Developer tools
I’d like to mention a few developer tools that I’m excited to experiment with:
- Jules
- Firebase Studio
- ADK
- Android XR
In-Person Experience
As a GDE and early tester for many Google AI products, I was showered with swags, in-person learning, community bonding and professional networking opportunities at I/O. I’m including a few photos to remember those moments.
The afternoon before I/O, there was a Building with AI workshop on AI agents with Gemini and ADK, led by Christina Lin and her team. Then in the evening there was a Pre-IO Happy Hour for GDEs (Google Developer Expert).
As a GDE, we had reserved front row seats at the Keynote.
I was invited to an exclusive Gemini Early Access Reception. It was great to meet the Gemini team and other early access testers. I even got a chance to play PingPong with GDE friend Vikram and two other Googlers.
Participated in gDAB feedback session on a advisory board for Developer X senior leadership.
AI Gathering event where we gained insights on the latest in AI from Google and other experts. It’s a gathering where we hear other start-ups pitch there ideas.
I participated in two video recordings at Google I/O: one is a long video for my I/O experience at the GDE Summit, and the other one is a YouTube Short part 2 for the official Google Cloud Tech channel, following Fashion with Imagen and Veo: Part 1. Both are to be released soon.
There are so many new models, tools and SDKs to try from I/O 2025: in Gemini app, Google AI Studio and Google Cloud Vertex AI. I look forward to sharing more in-depth tutorials soon!
