This is a submission for the Google I/O Writing Challenge

Google I/O 2026 showed us something that goes beyond a list of launches: a vision of where technology is heading.

Sundar Pichai (Google CEO) opened the presentation sharing some interesting numbers about the *evolution of AI with Google statistics that you can see in the following infographic.

But let’s get back to the event…
Over two hours, Google announced a wide range of products, updates and platforms. From new AI models to smart glasses, from music generation tools to a digital twin of the entire planet.
What stands out is not any single product on its own, but the way they are designed to work together. Most of them are not meant to exist in isolation but to integrate with each other and with the models Google is deploying across its ecosystem.

Below you will find all the launches organized by category, with my take on each one and direct links to the exact moment in the keynote where it was announced.

🧠 Models & Infrastructure

Behind almost everything presented at Google I/O 2026 has the same foundation: more advanced models and an infrastructure designed to scale new forms of interaction between people and systems.

Gemini Omni

The first model that allows you to modify videos using natural language, with inputs that can be images, text or video. But it is not just about understanding text, images, audio and video at the same time; it is about reasoning over all of them together to generate something new.
What sets it apart from any previous video generator is that it combines an intuitive understanding of physics with real knowledge about history, science and cultural context. So you can take a video you recorded and ask it to change what happens in it, edit the action, add characters, transform a moment into something completely unexpected.

📍 Gemini Omni → 17:15

Gemini 3.5 Flash

Gemini 3.5 Flash is the direct evolution of the main line of large language models (LLM) from Google, optimized to deliver ultra-high speed performance, advanced logical reasoning capabilities and code orchestration. It is an ideal model for building autonomous agents capable of executing complex task flows in the background, writing code, processing large text contexts or powering searches.
All of this while being four times faster than comparable models, which makes it the option specifically designed for agentic tasks.

📍 Gemini 3.5 Flash → 23:43

🤖 Agents & Productivity: Your Digital Life, Managed

This is one of the categories I enjoyed the most, featuring several tools that change the way we work and organize our daily tasks.

Gemini Spark

Gemini Spark is a personal AI agent that runs tasks in the background, across all your applications, without you having to supervise every step. With Spark you can organize an event, manage a chain of emails, coordinate a complex task across multiple services.
It also connects with external tools through the open MCP protocol, which extends its reach beyond the Google ecosystem.

📍 Gemini Spark → 35:18

Daily Brief

Google has been offering AI summaries for a while, but Daily Brief is something different. Instead of summarizing a document you provide, it reads your chats, your Gmail emails, your calendar context and your pending tasks, and then prioritizes what matters for your day. The difference between a generic summary and one that actually understands your context is significant, and that is exactly what Daily Brief proposes.

📍 Daily Brief → 1:15:06

Ask YouTube

Ask YouTube changes the search bar we know on YouTube into a conversation. You can ask for a summary, a specific recommendation, or request to find exactly what you need at a precise moment without watching the full video.
What I find most interesting is the impact on creators. The algorithm can now understand the deeper context of content and recommend videos that used to stay hidden behind generic titles. For creators, this is an opportunity, and for content consumption in general, it changes the way we interact with the platform.

📍 Ask YouTube → 7:41

Docs Live

Docs Live changes the way we create documents. While it's already possible to create content using Gemini's voice input options, this solution lets you share your ideas aloud, and Gemini will start creating a document, formatting, structuring, and writing the text in real time.
The key difference from Gemini's existing voice input is that the result isn't a chat response; it's a properly formatted Google Docs document, complete with headings, lists, and a professional structure right from the start.

📍 Docs Live → 9:14

🔍 Search & Commerce: Your Next Purchase Will Be Made by an Agent

AI search is already part of our daily routine, but this year Google took it further, turning search from a simple query into an agent that can look for information and even make purchases on your behalf.

AI Search Box

The new search box is no longer limited to text. It now accepts images, files, videos and even Chrome tabs as input. It may seem like a small change, but it completely transforms the experience of searching for something, powered by the new Gemini 3.5 Flash models.

📍 AI Search Box → 46:05

Search Agents

Search agents are a feature that, in my opinion, will become essential without many people noticing at first. They are background agents that you can set up to monitor specific topics, such as the value of a stock you are tracking, a flight route for an upcoming trip, or a property you want to rent in a specific neighborhood.

If the agent detects changes, it can summarize the information and notify you. It can pick up updates from blogs, news sites, social media, and real-time data on finance, shopping and sports.
This changes the current experience we have with Google Alerts, which were based only on keywords, since it takes both alerts and information retrieval to a much more advanced level.

📍 Search Agents → 47:48

Generative UI in Search

This announcement is quite interesting because information can already be accessed through agents, chats or other channels, but the main idea is to provide an interface that is intuitive for users to interpret that information. Instead of returning a list of links, Search can now build a personalized interactive interface for complex queries. Here you can get a live comparison table, a dynamic chart or an interactive explanation, all generated in real time for your specific question.

📍 Generative UI → 51:17

Universal Cart & UCP + AP2

Google is redefining the online shopping experience with two announcements. On one hand, Universal Cart turns the shopping cart into something where you can add products, and the system then works in the background autonomously, monitoring price drops, analyzing price history and notifying you when an item becomes available again.

On the other hand, the Universal Commerce Protocol (UCP) establishes an open standard that allows all of this to scale beyond Google. It creates a common language for agents and systems to work together across the entire shopping process, from finding a product to post-purchase support, connecting consumer platforms, businesses and payment providers.

UCP is compatible with other key ecosystem protocols such as Agent2Agent (A2A), Agent Payments Protocol (AP2) and Model Context Protocol (MCP), positioning it not as a Google tool, but as the infrastructure for agentic commerce.

📍 AP2 Protocol → 1:01:17
📍 Universal Cart → 1:03:13

🎬 Creative Tools: Can anyone make a movie today?

Content creation was another area where the shift in focus at Google I/O 2026 became very clear. It is not just about generating images, music or video, but about how these tools are starting to integrate into a continuous creative flow, closer to a conversation than to a technical process.

Google Flow

Google Flow is a creative platform developed by Google that allows users to generate, edit and compose videos, images and music from prompts or images. In its latest update, it integrates with Gemini Omni to take video editing to a more conversational level, allowing you to change environments, add characters, and generate 16 different camera angles from a single image.

📍 Google Flow → 1:28:08

Flow Music

Flow Music is a generative tool that allows users to compose songs and create full music videos from prompts. It also lets you provide a reference recording and build a complete track around it, edit it section by section, reimagine the style of a song while keeping its original melody, or create music videos by directly conversing with the agent.
I think it is an ideal tool for independent artists who develop games, apps or video content and want to create using AI.

📍 Flow Music → 1:31:01

Stitch

Google Stitch is a tool developed by Google Labs that uses AI to design user interfaces (UI/UX). These designs can be exported directly to code, Figma, Google Antigravity or Google AI Studio. The design process is driven by instructions that can be given through text or voice, and it is generated in real time.

📍 Stitch → 1:25:36

Google Pics

Google Pics is a new image creation and editing tool based on Nano Banana, Google’s model for this type of task. It allows users to select and edit specific elements with precision, such as moving objects, changing colors, or transforming one element into another without affecting the rest of the image.

Without a doubt, it is a very useful tool for content creators, making it easier to modify and edit images using text instructions.

📍 Google Pics → 1:24:20

⚙️ Developer Tools & Hardware

This section is perhaps one of the most diverse of the event, as it combines developer tools with consumer hardware that is still in active development. It ranges from systems capable of coordinating code agents at scale to devices that start bringing Gemini interactions directly into the physical world.

Antigravity 2.0

Antigravity 2.0 is a native desktop application that works as a central platform for coordinating multiple sub-agents running tasks in parallel. The keynote demo showed one of the most complex examples, building an operating system from scratch, and this tool allows you to create a plan and define how sub-agents should run in parallel in order to achieve the goal.

📍 Antigravity 2.0 → 26:55

CodeMender

CodeMender is a security tool originally developed by Google DeepMind. The tool scans code, identifies vulnerabilities autonomously, recommends fixes, tests them in a safe environment, and can apply the necessary patches with your approval at each step.
📍 CodeMender → 1:45:58

Audio Glasses

Audio glasses, developed with Samsung and designed in collaboration with Gentle Monster and Warby Parker, allow you to use Gemini without a screen and without taking your phone out. With these glasses, you can ask about a restaurant you just passed, get step-by-step directions, manage calls and messages, take photos with a voice command, or use the apps installed on your phone.
📍 Audio Glasses → 1:34:32

Display Glasses & Android XR

Display Glasses go one step further: they include micro-projectors built into the lenses that overlay useful information on the real world, such as navigation maps or real-time translations on signs, among other features.
Meanwhile, Android XR is the operating system platform that powers these devices, developed with Samsung and Qualcomm. It is still in a trusted testers phase, with a wider rollout expected later this year.
📍 Android XR / Display Glasses → 1:33:15

🔬 Science: What Happens When AI Never Stops Researching?

This was, for me, the most important section of the event. The following initiatives from Google apply AI to problems that go far beyond personal productivity, from accelerating scientific research to modeling the global climate and rethinking the process of drug discovery.

Gemini for Science

Gemini for Science is a research acceleration platform that allows scientists to stay up to date with newly published papers, turn research goals into executable code, and generate new hypotheses. It is still in a prototype phase in Google Labs, but the concept is what matters: AI is not presented as a replacement for scientific thinking, but as infrastructure that removes friction from the early stages of research, enabling literature search, synthesis of papers, and translation of hypotheses into experiments.
A researcher who can stay updated in real time across their entire field and automatically translate a hypothesis into an experiment is a researcher who can focus on more meaningful work.

📍 Gemini for Science → 1:46:37

AlphaEarth Foundations & WeatherNext

AlphaEarth Foundations is a model developed by Google DeepMind that works as an interactive digital twin of the Earth, powered by real-time satellite data, climate sensors, ocean readings, and biodiversity records. Meanwhile, WeatherNext is its atmospheric counterpart, a weather forecasting engine capable of predicting hurricane paths and extreme weather events with greater accuracy and speed than traditional systems.
These models were presented at the conference as examples of how Google’s AI technology can be applied to solve problems that affect billions of people around the world.

📍 AlphaEarth + WeatherNext → 1:46:37

Isomorphic Labs

Isomorphic Labs, the biotechnology company within Alphabet (a sister company of Google DeepMind), continues to build on AlphaFold. This is an AI system developed by DeepMind that enables the prediction of protein structures.
It is another example of how Google’s technology is being applied in the pharmaceutical industry, helping to significantly accelerate research and molecular design for treatments against cancer and complex immune disorders.
In the keynote, this work was described as “science at digital speed”, where AI acts as a tool to understand biological systems that were previously impossible to model directly.

📍 Isomorphic Labs → 1:46:37

SynthID

SynthID is Google’s invisible watermarking system for AI-generated content. This label is added to images, videos and audio at the moment of creation, allowing anyone or any system to later verify whether something was generated by AI.
This announcement is important not only from a safety perspective, but also because it is being adopted as a standard by companies such as OpenAI and ElevenLabs.

📍 SynthID → 21:06

What I'm Most Excited to Try

If I had to choose the three announcements I will follow most closely:

🛒 Universal Cart & UCP + AP2

These two announcements will change the way we shop online. For years, we have designed experiences for human users: visual interfaces, marketplaces, recommendations and conversion funnels. But Google is proposing something different: agents capable of discovering products, evaluating options, monitoring prices and executing purchases on our behalf.
This means ecommerce is no longer only a human platform interaction, but starts to become an ecosystem where agents also participate as consumers.
I do not think these solutions will replace traditional commerce overnight, but they will deeply change how trust is built, how products are presented, and how companies compete for attention, not only from people but also from intelligent agents.

🔍 Search Agents

Traditional alerts have always been passive: they depended on exact keywords and often generated more noise than context. This is one of the announcements I will probably follow most closely because it completely changes that logic.
Instead of manually searching for information, you can now delegate the monitoring of a topic to a system that understands intent, relevance and meaningful changes. An agent that continuously tracks the internet in the background.
And the more I think about it, the clearer it becomes that this might be one of the most important features of the keynote, precisely because it will quietly integrate into our daily routine.

🧬 Gemini for Science

Of everything announced, this is probably the project with the deepest potential impact.
Modern scientific research has a silent problem: the speed of knowledge has already surpassed human capacity to absorb it. Thousands of papers are published every week, information is fragmented, hypotheses are scattered, and entire weeks are spent just trying to stay updated.
Gemini for Science proposes something fundamentally different: turning AI into infrastructure for research.
The ability to translate scientific literature into actionable hypotheses, generate experimental code, connect discoveries across disciplines and accelerate research processes could completely change the scale at which science progresses.
Because perhaps the most important application of artificial intelligence is not to automate work, but to accelerate human knowledge.

🎥 What I Actually Tried: Google Flow

Most of the announcements that caught my attention were related to systems, agents and infrastructure. But beyond the long term vision, I also wanted to understand what it actually feels like to interact with one of these tools in practice.
So instead of just describing them, I decided to try one myself. I opened Google Flow and started experimenting with a short video prompt, and for a moment, I found myself living that childhood idea of creating my own animation.

While doing this, I could not help but think about how tools like this could evolve beyond individual experimentation. In the future, they might become part of how schools teach storytelling, narration and creative thinking, helping children express ideas through visual and generative tools. At the same time, it also raises an interesting challenge: how younger generations will learn to use these systems in a way that is both creative and intentional, rather than just consumptive.

What the Experience Felt Like

The result was an 8 second clip and honestly, the quality surprised me. With a fairly simple prompt describing a scene, the model generated something that looked cinematic and coherent. The prompt itself was also AI generated, and I am leaving it in the appendix at the end of this article in case you want to replicate the experiment.
A few things stood out from the experience.
The 8 second limit felt a bit frustrating at first. But after thinking about it, I am not sure if that is a limitation of Flow or simply how professional video production works. Scenes in film and TV are often short clips that are later assembled in post production. Flow seems to follow that same logic, where you build a story by connecting multiple clips instead of generating one long video in a single shot.

In a previous test, I was able to concatenate two clips, and what impressed me the most was the character consistency between them. The same character appeared in both scenes without drifting, keeping the same look and style. That is actually one of the hardest problems in AI video generation today: maintaining character consistency across different prompts. Flow’s approach of letting you define and save a character that can later be reused across scenes feels like a real step forward. Whether it can maintain that consistency across longer or more complex sequences is something I still want to keep testing.

How I Tested It

Under the hood, I used Gemini Omni Flash as the base model, with an 8 second duration, 2x speed and a 16:9 aspect ratio.
Each generation costs 50 credits, which gives a more concrete sense of the cost per clip when planning a longer project. The tool also allows direct publishing to YouTube and lets you select a custom thumbnail, making the end to end workflow surprisingly complete for a creative platform.

🔮 How far do we want to go?

Google I/O 2026 presented a series of solutions, models and use cases where artificial intelligence is becoming more deeply integrated into our daily lives. But while watching each demo, one broader question kept coming to my mind:

How far do we really want to go?

Because many of these tools do not only automate tasks, they also start to reorganize how we access information, how we make decisions and how we interact with the digital world.

What stood out the most was not each announcement on its own, but what happens when you look at them together. Agents that run in the background, search that becomes conversational, interfaces that adapt to context in real time. Not as isolated products, but as parts of a system that is still being built.

And in any system, what matters is not only each component, but how they connect with each other and what kind of emerging behavior appears when they start interacting.

Google showed the pieces based on AI.

What is still being defined is the full system: its direction, its shape, and how it will integrate into our daily lives and work.
And maybe that is the most important part of this new era: it is not only about what technology is capable of building, but about how each of us decides to interpret it, use it, and become part of it.

⚠️ A note on process and transparency

I am an organizer at GDG Barcelona, and I led the event where we brought together 37 people to watch the full Google I/O 2026 keynote live. That experience, I was able to follow the announcements in real time, listening to the reactions in the room ... is what shaped the perspective and opinions in this article.

This article is also based on the official sources listed in the references section, which I read and consulted directly after the event to verify and expand on each announcement.

All the visuals in this article were designed by me using Figma Design, because creating my own images is something I genuinely enjoy as part of my writing process. AI tools were used for text correction and translation assistance and all opinions, analysis and perspectives are my own.

📚 References

• Google. (n.d.). Gemini Omni. Google Blog. https://blog.google/intl/es-es/productos/presentamos-gemini-omni/ (Accessed May 22, 2026)
• Google. (n.d.). Gemini Spark. Gemini Overview. https://gemini.google/overview/agent/spark/ (Accessed May 22, 2026)
• Google. (n.d.). Daily Brief. Gemini Overview. https://gemini.google/overview/daily-brief/ (Accessed May 22, 2026)
• Google. (n.d.). The next evolution of the Gemini app. Google Blog. https://blog.google/innovation-and-ai/products/gemini-app/next-evolution-gemini-app/ (Accessed May 22, 2026)
• YouTube. (n.d.). YouTube News: Google I/O 2026. YouTube Blog. https://blog.youtube/news-and-events/youtube-news-google-io-2026/ (Accessed May 22, 2026)
• Google. (n.d.). YouTube search updates and AI features. Google Support. https://support.google.com/youtube/answer/16943763?hl=en (Accessed May 22, 2026)
• Google. (n.d.). Search at Google I/O 2026: AI-powered search updates. Google Blog. https://blog.google/products-and-platforms/products/search/search-io-2026/ (Accessed May 22, 2026)
• Google. (n.d.). Universal Cart and shopping updates. Google Blog. https://blog.google/intl/es-419/actualizaciones-de-producto/informacion/google-shopping-cart/ (Accessed May 22, 2026)
• Google. (n.d.). Merchant Center help: Shopping updates. Google Support. https://support.google.com/merchants/answer/16837055?hl=es (Accessed May 22, 2026)
• Google. (n.d.). Flow updates. Google Blog. https://blog.google/intl/es-419/actualizaciones-de-producto/flow-updates/ (Accessed May 22, 2026)
• Google. (n.d.). Stitch updates from Google Labs. Google Blog. https://blog.google/innovation-and-ai/models-and-research/google-labs/stitch-updates/ (Accessed May 22, 2026)
• Google. (n.d.). Google I/O 2026 developer tools highlights. Google Blog. https://blog.google/innovation-and-ai/technology/developers-tools/google-io-2026-collection/ (Accessed May 22, 2026)
• Google. (n.d.). Workspace updates at Google I/O 2026. Google Blog. https://blog.google/products-and-platforms/products/workspace/workspace-updates/ (Accessed May 22, 2026)
• Google. (n.d.). Antigravity 2.0. https://antigravity.google/product/antigravity-2 (Accessed May 22, 2026)
• Google. (n.d.). Developer highlights from Google I/O 2026. Google Blog. https://blog.google/innovation-and-ai/technology/developers-tools/google-io-2026-developer-highlights/ (Accessed May 22, 2026)
• Google Cloud. (n.d.). Innovations from Google I/O 2026 on Google Cloud. https://cloud.google.com/blog/products/ai-machine-learning/innovations-from-google-io-26-on-google-cloud (Accessed May 22, 2026)
• Google. (n.d.). Android XR at Google I/O 2026. Google Blog. https://blog.google/products-and-platforms/platforms/android/android-xr-io-2026/ (Accessed May 22, 2026)
• DeepMind. (n.d.). SynthID. https://deepmind.google/models/synthid/ (Accessed May 22, 2026)

📄 Appendix

This is the AI-generated prompt I used to create the video in Google Flow. I am including it here so you can replicate the experiment or explore it further.

A short animated intro video, 15 seconds. Chibi anime art style — soft cel-shading, vibrant neon colors, cinematic lighting, dark moody atmosphere. Think Lo-Fi anime meets cyberpunk gamer aesthetic.
Character: A small chubby panda in chibi style. Oversized black hoodie with hood down while walking, round panda ears visible on top. Serious and unbothered expression. Tiny paws. Soft black and white fur with subtle neon light reflections. This is a developer panda — cool, focused, says nothing.
Scene 1 — 0:00 to 0:05: Wide shot of a dark misty forest at night. A distant neon city skyline glows purple and cyan through the trees. Fog rolls along the ground. The panda walks alone from the right side of frame through a forest path, hands in hoodie pocket, completely unbothered. He approaches a large mossy rock formation — a hidden cave entrance covered by hanging vines with faint bioluminescent blue glow. He pushes the vines aside and steps in. Slow cinematic cut to black.
Scene 2 — 0:05 to 0:10: Interior of the cave — full gamer setup. RGB neon strips in Google colors (blue, red, yellow, green) line the rocky cave walls casting dramatic colored light on everything. A dark stone desk holds a glowing MacBook Pro, mechanical keyboard with RGB backlighting, mouse with neon underglow, and stacked empty energy drink cans. The panda walks to the desk, drops his backpack on the floor. Pulls out the chair and sits down. He opens the MacBook — a burst of white light floods his face and the cave. He slowly reaches to the side, picks up thick black sunglasses and puts them on. Then places large black headphones over his panda ears. He cracks his tiny paw knuckles. Leans forward. The RGB strips pulse once in sync.
Scene 3 — 0:10 to 0:15: Ultra slow cinematic push-in toward the MacBook screen. The cave darkens around it. The screen fills the entire frame glowing bright. Bold text appears: "GOOGLE I/O 2026 — THE AGENTIC ERA IS HERE" in clean white typography on dark background. Neon blue and green light pulses around the text edges. Below it, six glowing color blocks appear one by one with smooth fade-ins: MODELS · AGENTS · SEARCH · CREATIVE · DEV · SCIENCE. Each block in its Google neon color, white bold uppercase text, subtle neon glow border. Final frame holds 2 seconds with all blocks visible, neon pulsing softly. Fade to black.
Lighting & mood throughout: Dark, moody, cinematic. Neon reflections on all surfaces — the panda's fur, the cave walls, the desk. Color palette: deep black backgrounds, cyan #00F5FF, neon green #39FF14, Google blue #4285F4, Google red #EA4335, Google yellow #FBBC04, purple #BF5FFF. Inspired by cyberpunk anime aesthetics — think lo-fi coder vibes meets Akira color palette.

`