How Boris Cherny manages 2,000 overnight AI agents from a smartphone

Boris Cherni, creator of Claude Code, revealed details of his AI agent management system. It turns out that every night he launches about two thousand autonomous programs that independently write code, test, and improve the product.

Scale of Automation

Cherni developed a system that allows him to manage a huge number of AI agents with minimal attention overhead. All these agents work at night, while developers sleep, and perform routine tasks of writing and refactoring code. According to him, he manages this army primarily through his smartphone — sending commands, tracking progress, and adjusting the direction of work for the next day.

This is not just a demonstration of Claude's technical capabilities, but a practical workflow that works in reality. Cherni uses Claude Code as a platform for creating his own workflow. Agents work in parallel, independently of each other, and complete their assigned tasks without human intervention in intermediate stages.

The distinctive feature of his approach is that quality control remains in his hands. In the morning, Cherni reviews the results of the agents' work, accepts or rejects changes, and gives instructions for the next cycle. It's like managing a remote team, but at the scale of thousands of simultaneous 'employees'.

How It Works in Practice

This approach demonstrates how software development might look in the near future. Instead of hiring more programmers to perform routine tasks, a company can scale through automation. AI agents take on repetitive and predictable work:

• Refactoring existing code and improving readability
• Writing boilerplate code and standard components
• Finding and fixing potential bugs
• Optimizing performance and memory usage
• Generating unit tests and technical documentation

At the same time, control remains in human hands — Cherni monitors quality, checks the logic of changes, and guides agents in the right direction. This is critically important because AI can make mistakes in architectural decisions or miss the nuances of business requirements. The developer acts as an architect and quality integrator, not an executor.

Scaling Without Hiring New Developers

For startups and companies, this has obvious economic sense. Instead of hiring junior developers and spending months training them, you can give the tool to experienced specialists and allow them to focus on complex tasks. Simple and repetitive work will be solved by AI.

This is not about layoffs. History shows the opposite: tools expand a team's capabilities, allowing fewer people to do more. A developer equipped with 2000 autonomous agents can maintain a project that previously required a team of 20-30 people.

But there's an important nuance: this works only because an experienced developer stands behind the system, who knows what tasks need to be automated, how to formulate them correctly for AI, and what result is acceptable. Without such experience, the system will simply start generating low-quality code.

What This Means

Cherni's story is not about AI replacing developers. It's about how a quality tool in the hands of a professional can turn him into a developer with superpowers. Two thousand night agents are a scaling effect, not a replacement of labor. The previous generation of tools (IDEs, frameworks, cloud) already did the same. One specialist can now do the work that 20 years ago required a small team. AI is the next step in the evolution of development tools. For companies, the conclusion is simple: invest in tools for the best developers, not in the number of developers.