Best Practices for AI Documentation

Retrieval-Augmented Generation (RAG) systems, like Kapa, improve through high-quality documentation, benefiting both human readers and AI.

AI systems process documentation in chunks, relying on content matching and explicitly stated information.

For AI-friendly documentation, content must be explicit, self-contained, and contextually complete.

Chunking is necessary due to token limits and helps in targeted content retrieval, enhancing model comprehension.

Quick tips include using standardized semantic HTML, avoiding PDFs, and providing text equivalents for visuals.

Content organized hierarchically aids AI in maintaining context and extracting relevant information accurately.