Claude Code Update is INSANE
Claude Code has launched a web and mobile interface at claude.ai/code, allowing developers to kick off and monitor coding tasks remotely without a terminal. The update introduces cloud-based VM execution, persistent sessions, parallel task running, and an auto-fix feature for pull requests. This transforms Claude Code from a terminal-only tool into an async coding collaborator.
Summary
The video covers Anthropic's major update to Claude Code, which previously required users to be present at a terminal. The new experience is accessible via claude.ai/code and allows developers to connect their GitHub account, write a plain-English prompt, and launch full coding tasks directly from a browser. Anthropic runs these tasks on isolated cloud virtual machines pre-installed with Python, Node.js, Ruby, Go, Rust, Java, PostgreSQL, Redis, Docker, and Git. Claude clones the target repository, performs the work, and presents a reviewable diff in the browser.
A key capability highlighted is session persistence — even if the user closes their browser tab, Claude continues working in the cloud. Users can monitor and steer sessions from the Claude mobile app, answering Claude's questions or approving actions without being at a laptop. The presenter outlines several workflows: launching tasks directly from the browser, using a `--remote` flag in the CLI to offload work to the cloud while continuing local development, and a plan-then-execute workflow where Claude proposes a plan locally before running autonomously in the cloud.
The auto-fix feature for pull requests is described as particularly underrated. Once a PR is created, Claude monitors it on GitHub, automatically investigating and pushing fixes when CI checks fail, and addressing reviewer comments — with transparent disclosure in the thread that responses came from Claude Code.
On the infrastructure side, each session runs on fresh Anthropic-managed VMs with environment caching, so setup scripts only run once and future sessions boot from a snapshot. Security is handled via a proxy with scoped credentials, ensuring Git keys never exist inside the sandbox. Users can run multiple parallel cloud sessions simultaneously and use a 'teleport' command to pull a finished cloud session back into the local terminal with full conversation history intact.
Practical tips include maintaining a claude.md file in the repo for persistent context, using inline diff comments for cleaner review feedback, and monitoring context window usage with built-in commands. The video closes with availability information — Claude Code on the web is in research preview for Pro, Max, Team, and Enterprise users — and promotional mentions of the presenter's AI community and resource hub.
Key Insights
- The presenter explains that Claude Code on the web runs in Anthropic-managed isolated virtual machines pre-installed with a full development stack including Python, Node.js, Ruby, Go, Rust, Java, PostgreSQL, Redis, Docker, and Git — meaning users need nothing installed locally.
- The presenter highlights that cloud sessions persist even after closing the browser tab, allowing Claude to continue working autonomously and enabling users to check in and steer progress from the Claude mobile app.
- The presenter describes a plan-then-execute workflow using permission mode plan, where Claude explores the codebase and proposes a plan without touching code, which the user reviews before sending execution to the cloud — comparing it to a contractor reviewing blueprints before breaking ground.
- The presenter calls auto-fix for pull requests 'the most underrated feature,' explaining that Claude monitors a PR on GitHub, pushes fixes when CI checks fail, and addresses reviewer comments — while transparently labeling its own responses as coming from Claude Code so team members can distinguish agent output from human output.
- The presenter notes that users can run multiple remote sessions simultaneously — for example, fixing a bug on one branch, writing test coverage on another, and refactoring a module on a third all at once — and use a 'teleport' command to pull any finished cloud session back into the local terminal with full conversation history.
Topics
Transcript
[0:00] Claude Code update is insane. You're sitting at your desk, stuck, waiting for your dev environment to load. Again, you open your terminal, you type, you wait, you lose your train of thought, you switch windows five times just to get one thing done. Sound familiar? This is what most people's coding setup looks like in 2026, and it's costing you hours every single week. But what if you could kick off a coding task from your phone while you're getting coffee, and come back to a finished pull request? That's exactly what Claude Code on the web and mobile lets you do right now. And most [0:31] people have no idea it exists. Stay with me, because by…
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.