Building Claude Code with Boris Cherny

2026-02-25 · Source: The Pragmatic Engineer · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, medium

Summary

Boris Cherny, Head of Claude Code at Anthropic and former Principal Engineer at Meta, discusses the development and implications of Claude Code in a recent episode of *Pragmatic Engineer*. The discussion covers how Claude Code evolved from an internal side project into a core tool at Anthropic, enabling engineers to ship 20-30 pull requests daily by using five parallel Claude instances. Cherny highlights that a "good plan" allows Claude to "one-shot the implementation almost every time." The episode also details the architecture of Claude Code, including its "agentic search" which leverages glob and grep over more complex RAG solutions, and the rapid development of Claude Cowork in approximately 10 days. Cherny observes a shift in engineering roles towards generalism and rapid context-switching, likening current software engineers to medieval scribes facing an expanding market for written work.

Key takeaway

For AI Architects and VP of Engineering considering integrating AI into development workflows, understand that AI coding assistants like Claude Code can drastically increase output by enabling parallel development and shifting focus to planning. Your teams should prioritize clear architectural plans and maintain clean codebases, as these factors significantly enhance AI model effectiveness. Embrace prototyping over traditional PRDs to accelerate feature development and adapt to the evolving role of engineers towards generalism and rapid context-switching.

Key insights

AI-driven coding tools like Claude Code are transforming engineering workflows, emphasizing planning and rapid iteration over manual coding.

Principles

• Code quality directly impacts engineering productivity.
• Simple search methods can outperform complex AI approaches.
• Prototyping is superior to static specifications for AI products.

Method

Engineers can achieve high throughput by running multiple parallel AI agents, iterating on a plan, and then allowing the AI to generate the implementation in a single pass.

In practice

• Use parallel AI agents for rapid code generation.
• Prioritize finishing code migrations to avoid confusing AI models.
• Automate repetitive code review comments with lint rules.

Topics

• Claude Code
• AI-assisted Development
• Anthropic
• Engineering Productivity
• AI Agent Workflow

Best for: AI Architect, CTO, VP of Engineering/Data, AI Engineer, Machine Learning Engineer, Software Engineer

Related on AIssential

• Claude Code Isn’t a Coding Tool. It’s Your Team’s New Workflow Engine. — AI coding tools excel as workflow engines, not just code generators.
• Eight years of wanting, three months of building with AI — AI excels at tedious coding tasks but struggles with high-level architectural design and subjective problem-solving.
• The creator of Claude Code just revealed his workflow, and developers are losing their minds — Orchestrating multiple smart AI agents with self-correction and verification loops dramatically boosts developer productivity.
• I Built My Friend a Free Personalised Fitness Plan Online — Without Writing a Single Line of Code — AI can rapidly generate and deploy functional web applications from detailed prompts.
• Claude Code 2.0 MASSIVE Upgrade! (Game Changer) — Claude Code's latest updates enhance developer productivity through improved scheduling, memory, and multi-agent capabilities.
• Claude Coded My Python Project for 3 Hours Alone — Claude Code operates as an autonomous agent, independently developing and debugging complex Python projects with minimal human intervention.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Pragmatic Engineer.