Tuesday, May 19, 2026

Google I/O 2026 => agentic AI

Google I/O '26 Keynote - YouTube

 Google I/O 2026 keynote in 35 minutes - YouTube

The Google I/O 2026 keynote highlighted a major shift toward agentic AI and personal assistant capabilities integrated across Google's ecosystem. Below are the key announcements and features discussed:

AI Assistants & Personal Agents

  • Gemini Spark (10:25): A new 24/7 AI agent designed to manage your digital life, handle long-running background tasks, and act on your behalf across applications.
  • Daily Brief (20:12): A personalized morning digest that synthesizes information from your email, calendar, and tasks to help you start your day.

Search & Productivity

  • Ask YouTube (0:37): Allows users to ask questions about video content, providing digestible summaries and jumping to relevant segments.
  • Docs Live (1:42): Enables voice-driven document creation and editing, allowing users to verbally dictate complex tasks and formatting.
  • Intelligent Search Box (13:25): A major update to the search bar allowing multimodal inputs (text, images, files, videos) and seamless AI search integration.
  • Universal Cart (16:46): An intelligent shopping assistant that tracks products, finds deals, and alerts you to price drops across different services.

Models & Creative Tools

  • Gemini 3.5 Flash (6:55): A new, highly efficient, and fast model optimized for frontier intelligence and agentic coding. It is now the default for the Gemini app.
  • Gemini Omni (3:34): A generation tool capable of creating and editing content—including video, audio, and images—from any input type.
  • Google Pix (24:08): A new creative workspace tool for easy image editing and design.
  • Google Flow (25:08): Updates to the Flow suite now allow agents to execute multiple, complex creative tasks at once, such as generating video variations from a single image.

Hardware & Integration

  • Audio Glasses (28:33): New wearable glasses that provide hands-free, private access to Gemini for navigation, calls, and daily tasks, expected this fall.
  • Gemini for Mac (21:57): A native desktop app that utilizes Anti-Gravity technology to perform complex tasks like email drafting by reading selected files and PDFs.

Safety & Science

  • Content Credentials & SynthID (5:25): Google is expanding its watermarking and provenance verification to ensure users can identify AI-generated content across the web.
  • Gemini for Science (33:18): A suite of AI tools dedicated to accelerating scientific research, simulation, and drug discovery via Isomorphic Labs.

AI: Build With Gemini XPRIZE - $2M in prizes

 Build With Gemini XPRIZE - $2M in prizes. Pick a problem worth solving.

 a $2 million global hackathon, built on Google’s Gemini stack

90 days, until Aug 17, 2026

Ideate: Pick a real problem worth solving. AI helps you scope it.
Build: Build a working prototype using AI.
Ship: Deploy the business. Real users in production.
Grow: Market it. Grow it. Show the revenue.

5 categories:
Half of global GDP runs through five sectors:
  • Education. 
  • Human potential. 
  • Entrepreneurship and small business. 
  • Financial access. 
  • Professional services.

offered resources are not exactly enough for anything meaningful... 
it is mostly advertising...

Build with Gemini
$100 Google AI Ultra Bonus: Claim in the Antigravity app. Activates seamlessly the moment you hit your standard quota limit. Redeem by May 25, 2026. Credits available until supplies last. Terms apply.
$300 Google Cloud Credit: Free credits for new customers to host and scale securely.





Example: 

High Agency Is the New Credential

A 41-year-old self-taught coder in LA named Matthew Gallagher launched a GLP-1 weight-loss telehealth platform in September 2024. He spent $20,000. An AI chatbot to handle customer service.

His team consists of two people: him and his brother.

In his first year revenue crushed $401 million with a total of 250,000 customers. His current run rate: over $3 million per day, tracking toward $1.8 billion in 2026.

I want you to think about that for a second. Two people. $20,000. A dozen AI tools wired into an existing supply chain. No venture round. No engineering team.

That is what happens when high agency meets a real, painful problem: the cost of building drops to nearly zero.

bun.js: Zig => Rust, with AI

Bun.js is a flagship user of new Zig language, that is "human friendly".

Rust is much bigger and more complicated language, and while human developers who know it like it,
it is not easy. But now with AI Agents being able to code, 
and Bun as part of Anthropic having likely unlimited access to AI,
a full translation from Zig to Rust apparently worked well enough, and quickly!

If nothing else this is an indication of what can and may be done with many other projects.
Not to mention Microsoft's even more controversial intent to port Windows apps to Rust from C++.

Rewrite Bun in Rust by Jarred-Sumner · Pull Request #30412 · oven-sh/bun @GitHub

It passes Bun's pre-existing test suite on all platforms (and fixes several memory leaks and flaky tests), the binary size shrinks by 3 MB - 8 MB, the benchmarks are between neutral and faster - and most importantly, we now have compiler-assisted tools for catching & preventing memory bugs, which have costed the team an enormous amount of development & debugging time over the years.
The codebase is otherwise largely the same. The same architecture, the same data structures. Bun still uses few 3rd party libraries. No async rust.

Bun Rewrites 1M Lines of Zig to Rust in 6 Days


Bun’s Bold Gamble: Porting a Million Lines from Zig to Rust with AI | by Sahil | May, 2026 | Medium



related

HyperFrames.js: HTML + AI => Videos

HyperFrames: Write HTML and JavaScript to Create Videos 

An open-source framework for creating and rendering videos with HTML and JavaScript. Essentially a simpler non-React alternative to Remotion

It includes a variety of built-in blocks/components for common video effects and elements, and can also composite existing video and audio clips. 


Hyperframes is an open-source video rendering framework that lets you create, preview, and render HTML-based video compositions — with first-class support for AI agents.