When the creator of the world’s most advanced coding agent speaks, Silicon Valley doesn’t just listen — it takes notes.
For the past week, the engineering community has been dissecting one Thread on x from Boris CherryCreator and Head Claude Code at Anthropic. What began as a casual sharing of his personal terminal setup has turned into a viral manifesto on the future of software development, with industry insiders calling it a watershed moment for the startup.
"If you’re not reading cloud code best practices directly from its creator, you’re behind as a programmer." It is written Jeff Tanga prominent voice in the developer community. Kyle MacNeiceanother industry observer, and made this announcement with Cherry "Game-changing updates," is anthropic "fire," likely to be encountered "Their Chat GPT moment."
The excitement stems from a paradox: Cherry’s workflow is surprisingly simple, yet it allows a single person to operate with the production capacity of a small engineering department. As one user noted on X after implementing Cherry’s setup, the experience "Feels more like StarCraft" Compared to traditional coding – a shift from typing syntax to command independent units.
Here’s a workflow breakdown of how software is built, straight from the architect himself.
Running up to five AI agents simultaneously turns coding into a real-time strategy game
The most surprising revelation from Cherney’s revelation is that he does not code in a linear fashion. In the traditional "Inner loop" In development, a programmer writes one function, tests it, and moves on to the next. Cherry, however, serves as fleet commander.
"I run 5 clades in parallel in my terminal," Cherry wrote. "I number my tabs 1-5, and use system notifications to know when a cloud needs input."
Utilizing ITMR2 system information, Cherry effectively manages five simultaneous work streams. While one agent runs a test suite, another a legacy module, and a third a reactor drafts documents. He also walks "5-10 clods on Claude E" In his browser, using a "Teleport" Command to close the session between the web and its local machine.
It validates "Do more with less" The strategy was outlined earlier this week by Anthropic president Daniela Amodei. While competitors like Openai pursue multi-trillion dollar infrastructure buildouts, Anthropic is proving that higher orchestration of existing models can yield productivity gains.
A counter-argument to choosing the slowest, smartest model
In a surprising move for the industry of late, Cherry revealed that it exclusively uses Anthropic’s heaviest, slowest model. Ops 4.5.
"I use opus 4.5 with think for everything," cherry Explained. "It’s the best coding model I’ve ever used, and even though it’s bigger and slower than Sonnet, because you have to minimize it and it’s better at using the tool, in the end it’s much faster than using a smaller model."
For enterprise technology leaders, this is an important insight. The bottleneck in modern AI development is not the speed of token generation. This is human time spent correcting AI errors. Cherney’s workflow suggests that paying "Compute tax" A smart model ends this for the front "Tax reform" Later on
A shared file turns every AI mistake into a permanent lesson
Cherny also explained how his team tackles the problem of AI amnesia. Standard models of large languages do not "Remember" A company’s specific coding style or architectural decisions from session to session.
To indicate this, the Cherry team maintains a file name Claude.Mid In their git repository. "Whenever we see Claude doing something wrong we add it to Claude.md, so Claude knows not to do it next time," He wrote.
This practice turns the codebase into a self-correcting organism. When a human developer reviews a pull request and spots an error, they don’t just fix the code. They tag the AI to update their instructions. "Every mistake becomes a rulefor , for , for , ." Noted Akash Guptaa product leader in thread analytics. The longer the team works, the smarter the agent becomes.
Slash commands and subagents automate the most tedious parts of development
"Vanilla" Workflow One Observer is defined as driven by tight automation of repetitive tasks. CHERNY SLASH COMMANDS – Custom shortcuts are checked into the project repository to handle complex operations with a single keystroke.
He highlighted a command /push-pr commitwhich he calls dozens of times a day. Instead of manually typing git commands, writing commit messages, and opening pull requests, the agent handles the bureaucracy of version control autonomously.
Cherney has also deployed subagents — specialized AI personalities — to handle specific stages of the development lifecycle. He uses a code-simplifier to clean up the architecture after the central work and a verification app agent to run end-to-end tests before any ship.
Why Validation Loops Are the Real Unlock for AI-Infused Code
If there’s one reason why Cloud Code allegedly killed $1 billion in annual recurring revenue So quick, it’s a validation loop. AI is not just a text generator. This is a tester.
"Cloud checks every single change I make to cloud.e/code using the Cloud Chrome extension," Cherry wrote. "It opens a browser, tests the UI, and runs until the code works and the UX feels good."
He argues that giving AI a way to validate its work — whether through browser automation, running a baz command, or executing a test suite — improves the quality of the final results. "2-3x" An agent doesn’t just write code. It proves the code works.
What Cherry’s Workflow Hints About the Future of Software Engineering
The reaction to Cherney’s thread shows how developers think about their craft. for years, "AI Coding" Which means an automatic function in a text editor – a faster way to type. Cherry has demonstrated that it can now serve as an operating system for labor.
"If you’re already an engineer, read this … and want more power," Jeff Tang Summarized on X.
The tools to multiply human output by a factor of five are already here. All they need is a willingness to stop thinking of AI as an assistant and treat it as a workforce. Programmers who make the mental leap first will not be very productive. They will be playing a completely different game – and everyone else will still be typing.