Google Antigravity 2.0 Is INSANE!
The video introduces Google Antigravity 2.0, a rebuilt AI coding and task management tool unveiled at Google I/O on May 19th. The presenter highlights five major new features—sub-agents, asynchronous tasks, scheduled tasks, JSON hooks, and voice input—arguing it transforms from a coding tool into a team of autonomous AI workers. The video is also a promotional pitch for the presenter's paid community, the AI Profit Boardroom.
Summary
The video opens with a dramatic demonstration of Google Antigravity 2.0's capabilities: at a Google I/O showcase, the tool was given the task of building a complete computer system from scratch. To accomplish this, it autonomously split into 93 parallel sub-agents that collectively produced approximately 2.6 billion words of code. The presenter uses this as a hook to argue that Antigravity 2.0 has fundamentally outgrown its identity as a coding tool.
The presenter walks through five core new features. First, sub-agents: the main AI acts as a manager, breaking large tasks into smaller pieces and delegating each to a dedicated helper AI with its own isolated context window. This solves the problem of earlier AI models losing track of details on large jobs. Second, asynchronous tasks: the AI works in the background without requiring the user to monitor it, allowing users to do other work while tasks complete. Third, scheduled tasks: users can assign recurring jobs on a timer (daily, weekly, monthly), enabling automated morning briefings or weekly content plans with no manual trigger needed.
Fourth, JSON hooks: these are conditional rules ('when this happens, do that') that make the AI behave automatically based on predefined triggers, reducing the need for human oversight. Fifth, voice input: users can speak commands to the tool and see their words transcribed in real time, enabling on-the-go task delegation without typing.
The presenter contextualizes Antigravity 2.0 against the original version from November, describing the new version as running on Gemini 3.5 Flash—reportedly four times faster than its predecessor—and notes that Google is sunsetting its legacy command tool in June in favor of this platform. Throughout, the presenter frames the broader shift as a move from 'being the worker' to 'being the manager' of AI systems.
The video includes multiple promotional segments for the AI Profit Boardroom, a paid coaching community with 2,800 members, and a free alternative called AI Success Lab with 67,000 members. Both are presented as venues for hands-on implementation of tools like Antigravity 2.0.
Key Insights
- During a Google I/O demo, Antigravity 2.0 autonomously split itself into 93 parallel sub-agents that collectively wrote approximately 2.6 billion words of code to build a full computer system from scratch, illustrating its multi-agent orchestration capability.
- The presenter argues that the sub-agent architecture solves a core limitation of prior AI models: when given large tasks, a single AI would run out of effective working memory—described as 'a desk that gets too messy'—causing slowdowns and errors on complex jobs.
- Antigravity 2.0's scheduled tasks feature allows users to assign recurring automated jobs on a fixed timer, such as a daily 6 a.m. content briefing compiled before the user wakes up, removing the need for any manual trigger.
- The presenter states that Google is shutting down its old command-line tool in June and migrating all users to Antigravity 2.0, framing this as a company-wide strategic bet on the AI-worker-management paradigm rather than a minor product update.
- Antigravity 2.0 runs on Gemini 3.5 Flash, which the presenter claims produces output approximately four times faster than the model powering the original Antigravity released in November.
Topics
Transcript
[0:00] Google anti-gravity 2.0 is insane and it just stopped being a coding tool. Google put this thing on a stage and gave it one job. Build a whole computer system from scratch. Not a small app, a full system and it did it. Here's the part that got my attention. Didn't do the work alone. Split itself into 93 separate AI workers. All of them working at the same time on different pieces side by side. Together they wrote about 2.6 billion words of code before they were done. So let me say that again in plain English. One AI looked at a giant job, broke it to small pieces and handed [0:31] each piece to its own little…
Full transcript available for MurmurCast members
Sign Up to AccessMore from Julian Goldie SEO
NEW Nvidia Autonomous AI is WILD!! 🤯
Nvidia announced Nemo Clo, a new autonomous AI agent system that operates independently without continuous prompting. Powered by Nemotron 3 Ultra (a 550 billion parameter model), the system is five times faster and cheaper than previous versions, with OpenShell providing secure sandboxed execution.
Laguna XS 2.1: New FREE + Opensource Local AI!
Julian reviews Laguna XS 2.1, a new free open-source local AI coding model from Poolside that performs comparably to Qwen 3.6 and outperforms Claude Haiku on benchmarks. He demonstrates its practical capabilities by building landing pages and functional apps, highlighting its speed, offline functionality, and multiple deployment options through local setup, Claude Code, or OpenRouter's free API.
How to Run Hermes FREE Forever!
The video demonstrates how to run the Hermes AI agent for free using Gemma 4, a local open-source model from Google, with significant speed improvements through MLX optimization. The setup works on Apple Silicon Macs or via free APIs on Open Router, enabling autonomous agents to work offline and privately without subscription costs.
This NEW Chinese AI is INSANE! (FREE + Open Source!)
Long Cap 2.0 is a new open-source Chinese AI model from a food delivery app company that offers 1 million tokens of free context memory, beats GPT-4.5 on SWE bench pro benchmarks, and uses efficient parameter activation to reduce computational overhead while maintaining high performance.
Claude Code is now FREE: Here’s how…
Google's new Gemma 4 model running on Ollama is 90% faster on Apple Silicon, enabling free Claude Code usage locally without token costs. The setup requires three simple steps: downloading Ollama, Gemma 4, and installing into Claude Code, with alternatives available via OpenRouter API for non-Mac users.