PDF upload

Under Dashboard → Agent → Knowledge you can upload PDF documents that supplement site scraping. Only PDF MIME types are accepted; each file is read up to roughly three hundred twenty kilobytes for text extraction, and a longer excerpt (up to twelve thousand characters) can be stored for the AI context. You can keep up to fourteen PDFs per organization alongside your crawled site knowledge.

Long-form specs, security whitepapers, and one-pagers often hold detail your marketing pages skip. Adding them gives the agent grounded references for technical and compliance questions without pasting paragraphs into the tone or custom instructions fields.

Prefer text-based PDFs; scanned pages without a text layer extract poorly. Split huge decks into the chapters visitors actually ask about, and remove outdated versions when you ship replacements.

Use PDFs for stable reference material and keep time-sensitive numbers on the live site where you can update them and re-scrape.