If docs live inside WordPress - as a custom post type, regular posts, pages, or block patterns - SleekAI reads them natively without any export step. For docs hosted elsewhere (Mintlify, ReadMe, GitBook), the practical path is to mirror the Markdown into an OpenAI Files vector store. Either way the chatbot widget still lives on your WordPress site, where the rest of your marketing already runs.

Yes - the system prompt instructs the model to end every reply with a link to the doc page that informed it. For technical content this is non-negotiable: developers will not trust an unsourced claim about authentication or rate limits. Citations also create a feedback loop - if a link is dead or points to the wrong section, you fix the doc and the bot improves immediately on the next query.

Stuffing 800 endpoints into a single prompt is wasteful and inaccurate. Use OpenAI Files as a vector store - SleekAI retrieves only the endpoints relevant to the question, keeps token use predictable, and stays accurate even on huge references. Some teams split further: a dedicated 'webhooks bot' and 'billing bot' each scoped to a smaller corpus via multibot.

Yes - fenced code in your docs flows into the prompt with language hint and indentation preserved, so the bot can quote correct samples back without re-indenting them. JSON payloads, cURL one-liners, Python and Go SDK examples, even Bash exports all survive the round trip. Tables of error codes or rate limits are kept structured rather than flattened to prose.

Yes. Display conditions scope a chatbot to URL patterns, post types, or logged-in roles, and multibot lets you spin up several focused assistants - 'API docs bot', 'billing bot', 'admin bot' - each with its own scope and system prompt. A developer reading /docs/webhooks gets the webhooks bot; a finance lead in /docs/billing gets the billing bot.

SleekAI is a one-time WordPress plugin license, not a per-conversation SaaS subscription. You bring your own OpenAI API key, so you pay OpenAI directly at standard API rates - usually a fraction of what white-label docs chatbots charge. There is no token markup, no resolved-conversation fee, and no monthly seat cap.

Two levers. First, the system prompt explicitly forbids invented endpoints and tells the model to defer when context is missing. Second, retrieval keeps the model grounded - if the right doc is in context, hallucination drops sharply. Conversation logs let you spot any drift and tighten the prompt or fill the doc gap that caused it.

Yes - tag each SDK page or sample with a language hint and let the bot detect from the question whether the developer is asking about Python, Go, Node, or Ruby. The reply quotes the matching sample. Some teams prefer a separate bot per language for cleaner scoping; multibot makes that a config change rather than a rebuild.