Why write code in 2026
2 days ago
- #software-factory
- #coding-practices
- #ai-agents
- The role of software engineers is to build the 'software factory', including the infrastructure that enables agents to successfully prompt for changes and ship immediately.
- Proactive measures like prompts (skills, AGENTS.md, knowledge bases) and reactive protections through automated evaluation (tests, linting, type systems, evals) keep agents on track, allowing even simple models to produce good-enough changes.
- While AI-generated code is prevalent, writing code remains valuable for direct thinking in the execution environment, fostering attention, understanding, and ownership, rather than passively observing agents.
- Writing code helps experience system fragility and architectural principles, enabling cleanup, documentation, and debugging to improve the software factory and prevent sloppy code accumulation.
- Agents are compared to freshly onboarded interns, not compilers, as they work with partial or sloppy code and imprecise change descriptions, making human oversight crucial to avoid amplifying bad decisions.
- Hands-on coding allows humans to guide agents, delete unnecessary code, and explore better architectures, enhancing authorship and the ability to shape the software factory.
- Details matter in building a software factory, from architectural patterns to algorithms and performance; agents encourage evaluation, measurement, and early CI integration, improving software practices.
- Like in a car factory, occasionally dissecting the assembly line or delving into details (e.g., internal combustion engines or brake pads) is necessary to address weaknesses and connect minute details to the big picture without arbitrary boundaries.