Developers now write documentation for AI tools, transforming tech writing

Developers are increasingly writing and structuring documentation specifically for AI tools to consume, transforming technical writing into “context curation” as artificial intelligence systems require well-organized information to function effectively. This shift represents a fundamental change in how documentation is created and used, with technical writers positioned to become essential “context curators” who design information architectures that serve both human and AI needs.

What you should know: The rise of AI-powered development tools has made documentation quality directly impact code generation and system performance.

• Large language models (LLMs) require clear, accurate, and well-structured inputs to produce useful outputs, making the quality of documentation crucial for AI effectiveness.
• Developers now pay close attention to context window sizes—the amount of information that can be fed to an LLM—with newer models able to process entire book-length documents.
• Engineers are discovering that writing skills, previously considered “soft skills,” are now essential for successful AI-assisted development.

How context curation works: Technical writers are evolving into specialists who orchestrate content strategies around both human and AI requirements.

• Context curators create and maintain elaborate context folders that AI tools can access to build solutions autonomously with greater accuracy.
• The role involves understanding information architecture, semantic tagging, and documentation markup to optimize AI consumption.
• Tools like Claude Code recommend creating CLAUDE.md files containing project instructions and guidelines, representing just the beginning of structured AI documentation.

The bigger picture: This development represents a return to documentation-driven development, where writing becomes integral to the software creation process.

• Technical writers can now participate directly in API design and development, with their word selection skills becoming crucial for software functionality.
• The trend points toward “docs-as-data,” where documentation serves as modular components that can be inserted into AI-powered development environments.
• Most technical writing teams are already serving llms.txt files and LLM-optimized Markdown, but the field is moving toward more sophisticated semantic markup standards.

What industry experts are saying: Fabrizio Ferri Benedetti, the author, positions this as a natural evolution of technical writing’s core mission.

• “Context is so much better than content (a much abused word that means little) because it’s tied to meaning. Context is situational, relevant, necessarily limited. AI needs context to shape its thoughts,” he explains.
• Technical writers are described as “jacks of all trades that save the day” who can adapt to serve both human understanding and AI processing needs.

Looking ahead: The endgame involves making content accessible to both LLMs and humans, allowing each to extract knowledge tailored to their specific needs.

• Standards for AI-optimized documentation packaging don’t yet exist, but semantic markup languages like DITA may experience a resurgence.
• Future tools may include interfaces that let users download specific documentation versions and formats optimized for different AI systems.
• Technical writers wearing “art gallery curator hats” will guide both human and artificial intelligence through complex information landscapes.