PAX, an ERP software maker, is publishing its documentation in AI-friendly formats—raw Markdown files and indexed bundles—so AI assistants can answer product questions without forcing users to contact support.

What happened: PAX created two AI-accessible versions of its public documentation: individual pages in raw Markdown (accessed by appending .md to the normal URL) and two bundled files (llms.txt as an indexed map, and llms-full.txt as a complete documentation bundle). Internal links within AI-facing docs point to other Markdown pages, so AI systems stay within the cleaner format.

Why it matters: PAX's AI assistant, Paxy, can now answer workflow and product questions—such as how to receive a purchase order or configure shipping—directly from public docs rather than requiring users to file support tickets. This approach separates how-to help (which AI can provide) from sensitive business data access (which remains permission-scoped), letting small manufacturers get faster answers without human intervention.

What to watch: The documentation structure itself—clear titles, concise overviews, predictable sections, good related links, and accurate source structure—benefits both humans and AI. The company treats AI-friendly documentation as an extension of writing clear, well-organized help pages that serve any reader looking for accurate information quickly.