DHH’s new way of writing code

2026-04-01 · Source: The Pragmatic Engineer · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning, Entrepreneurship & Start-ups · Depth: Advanced, extended

Summary

David Heinemeier Hansson (DHH), co-founder of 37signals and creator of Ruby on Rails, discusses his dramatic shift to an "agent-first" AI workflow after initially being skeptical of AI coding tools. This change, largely driven by the capabilities of models like Opus 4.5 and agent harnesses, has significantly accelerated his productivity and ambition in software development. DHH highlights how AI agents, particularly when used with tools like OpenCode and Claude Code, enable rapid prototyping and complex project execution, such as developing the Umachi Linux distribution and a Basecamp CLI. He emphasizes that while AI accelerates code generation, human judgment, taste, and a high bar for code quality remain paramount, especially for senior developers who can effectively review and direct agent output. This shift is leading to an "exploding pie" of new projects and a re-evaluation of traditional software development methodologies and team structures.

Key takeaway

For AI Engineers and Software Architects evaluating new development paradigms, embrace an "agent-first" workflow to dramatically increase productivity and tackle previously unfeasible projects. Focus on honing your judgment and design sensibilities, as these human skills become even more critical for guiding AI agents and ensuring high-quality, maintainable software, rather than merely generating code.

Key insights

AI agents, particularly advanced models, are transforming software development by dramatically increasing individual productivity and project ambition.

Principles

• Aesthetics is truth in software design and engineering.
• Smaller, highly capable teams accelerate product development.
• The UNIX philosophy of small, interoperable tools is validated by AI agents.

Method

Adopt an "agent-first" workflow, starting new projects by prompting AI agents to generate initial drafts. Review and refine agent-produced code, leveraging senior expertise for quality assurance and complex problem-solving.

In practice

• Use agent harnesses (e.g., OpenCode) with frontier models like Opus for coding.
• Employ multiple AI models simultaneously for varied task speeds and critiques.
• Prioritize code quality and human review for agent-generated code in production.

Topics

• AI Agent Workflow
• Ruby on Rails
• 37signals
• Software Craftsmanship
• Developer Productivity

Best for: Software Engineer, AI Engineer, Director of AI/ML

Related on AIssential

• Coding Agents Meet Data Science — AI coding agents boost developer productivity but introduce new bottlenecks and require specialized adaptation for data science workflows.
• How frontier teams are reinventing AI-native development — AI-native development, by restructuring workflows around agents and providing rich context, dramatically boosts software delivery.
• How (and what) I'm building this week — Interactive, agent-driven learning and development workflows enhance building and deployment efficiency.
• Thoughts on coding agents — Coding agents fundamentally shift software engineering workflows, demanding explicit context provision and deliberate model-building from developers.
• Mitchell Hashimoto’s new way of writing code — AI agents are transforming software development workflows and challenging traditional open source and version control paradigms.
• Agentic Workflows Have Changed EVERYTHING in 2026 (DEATH Of The Senior Dev?) — Agentic AI workflows have fundamentally reshaped software development by 2026, making traditional coding skills obsolete while elevating engineering principles.

Open in AIssential →

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