Up to roughly a hundred posts can be sent inline as context per request, which is fine for most blogs. Beyond that, OpenAI Files vector storage indexes the full archive and retrieves only the relevant chunks per question, so a five-thousand-post archive performs the same as a fifty-post one. The cost difference is small, mostly the embedding step at index time.

Yes. Each answer can include a link back to the post the answer was drawn from, configurable per chatbot via the system prompt. Citations build reader trust and route traffic to the original article, which is critical if you monetize through ads, affiliates, or email signup. Readers verifying a claim is what makes them subscribe.

By default it only reads published, public content. Display conditions and post status filters let you fine-tune the scope further, for example excluding sponsored posts or a specific category. If you have private members-only posts, you can scope a separate chatbot for logged-in users only.

Yes. Any public custom post type can be added to the context, including portfolios, recipes, reviews, podcast episodes, or interview transcripts. The bot treats each type as a distinct source so an answer about a recipe links to the recipe page rather than a stray tag archive.

The system prompt is fully editable, so the bot can speak like your blog rather than like a generic assistant. Most blog owners spend ten minutes pasting in a few example answers and a voice description, then iterate from real conversations. The name, avatar, and accent colors are also configurable so it feels like part of the site.

If WPML or Polylang serves a translated version of the page, SleekAI loads context from that language's posts and replies in kind. A French reader on the French version of your site gets French answers drawn from French posts, with French citations. The system prompt can also be language-specific.

Yes. The bot reads code blocks as plain text and can quote them in answers, which is useful for technical blogs whose readers ask 'show me the snippet that does X'. If you want the formatting preserved, mention that in the system prompt and the bot will return fenced code blocks rather than inline text.

Conversation logs are stored in your WordPress database and are searchable from the admin. You can sort by frequency to find the top ten unanswered questions, export to CSV for analysis, or purge old logs on a schedule. Nothing is shared with Sleek or any third party beyond the OpenAI calls you control.