Accessible HTML, self-contained answers, crawlers and structured data: how to make a site easier to understand, without promising citations.
From search to a sourced answer
An assistant with web search can discover links and snippets, then consult selected pages to prepare a sourced answer. The path depends on the product, available tools and question; it does not necessarily retrieve every page in full. There is no universal documented rule requiring 150–400-word chunks or prioritising the first 30% of a page.
Make content available in HTML
Initial HTML makes content readable without running JavaScript. Googlebot can render JavaScript, however, so not all crawlers should be described as unable to do so. Our guides and descriptions are in HTML; JavaScript handles local PDF processing. Reading the documentation and executing a conversion are separate capabilities.
Separate search, browsing and training
OAI-SearchBot supports ChatGPT search; GPTBot concerns training. PerplexityBot supports Perplexity search. Their purposes differ. Allowing a crawler in robots.txt permits crawling under its rules, without guaranteeing indexing or citation. We retain the existing open policy; this article does not change crawler permissions.
Write answers that stand on their own
Start a section with a direct answer, then explain steps and exceptions. For a PDF tool, state formats, limits and conversion losses. For example, our export offers 72, 144 and 216 DPI; total input is limited to 100 MB, and an operation exports up to 100 pages. These describe the service, not an AI ranking mechanism.
Use metadata without promising visibility
An author, an update date and consistent BlogPosting markup make a page more explicit. Structured data must match visible content and cannot replace it. Google says no special schema or AI text file is required for its AI search features. A visible FAQ remains useful, but markup guarantees neither citations nor rich results.
What we publish for readers and agents
The site provides stable URLs, linked translations, a sitemap and an llms.txt catalog. That catalog is an optional entry point, not a conversion API or permission to access documents. No user files are published in articles, the sitemap or the catalog. Browser conversions stay local, without ads or tracking.
Frequently asked questions
Does HTML guarantee an AI citation?
No. Accessible HTML makes reading easier, but selection depends on the engine, query and available sources.
Does llms.txt allow PDF conversion over HTTP?
No. It lists public pages. Available conversions run in the browser; no HTTP conversion endpoint is provided.