Legible
Documents

How to Upload PDFs & Word Docs to Legible's Knowledge Layer

Learn how to easily upload PDFs, Word docs, and notes to Legible's knowledge layer. Enhance your AI chatbots with custom context for better, more informed answers.

6 min readUpdated 2026-03-21Chatbots & RAG
Why this matters

Legible's document upload feature lets you add extra knowledge directly into the content layer. This is ideal when important context lives in PDFs, Word documents, Markdown files, or text notes instead of on the public site.

Uploaded documents strengthen Content Chat and integrated assistant experiences by expanding the retrieval layer with more complete context.

What The Documents Tool Does

When you upload a file to Legible, the service extracts text, stores it as a document content item, and prepares it for retrieval. In practice, that means the content becomes part of the same AI knowledge layer used by Content Chat and the AI Export API.

This is useful for files that explain product details, policies, onboarding steps, implementation notes, or support knowledge that is not fully represented on the website.

Supported Use Cases

  • Upload PDFs and Word docs to improve support answers.
  • Add Markdown or text reference files to enrich the assistant's context.
  • Include content that is important for chat quality but not appropriate for public web pages.
  • Bring in product or operational context that would otherwise be missing from a website-only knowledge base.

How Documents Flow Through Legible

  • The product handles extraction and token counting for the uploaded file.
  • The document is chunked so it can participate in retrieval workflows.
  • This gives your AI assistant more context without requiring a separate ingestion pipeline.
Upload file to Legible
-> Legible extracts text
-> Legible stores it as a document content item
-> Legible chunks it for retrieval
-> Content Chat and integrated assistants can use that content as additional context

What Documents Are Best For

  • Reference material that improves answer quality.
  • Long-form help content that is awkward to place on the public website.
  • Operational or product explanations needed by support and success teams.
  • Supplementary knowledge that should inform answers even if it does not belong in the marketing site navigation.

What you do not have to build

  • No custom OCR, parsing, or file-ingestion system to build.
  • No separate chunking and retrieval prep for uploaded files.
  • Uploaded docs join the same knowledge layer as website content, FAQs, and AI-readable exports.
  • Teams can improve chatbot context quickly without restructuring the public website.