{"id":7463,"date":"2026-08-19T19:10:06","date_gmt":"2026-08-19T19:10:06","guid":{"rendered":"https:\/\/www.wiven.ai\/wiven-llm\/docs\/utilisation\/documents\/"},"modified":"2026-08-19T19:10:06","modified_gmt":"2026-08-19T19:10:06","slug":"documents","status":"publish","type":"wvn_doc","link":"https:\/\/www.wiven.ai\/en\/wiven-llm\/docs\/utilisation\/documents\/","title":{"rendered":"Documents"},"content":{"rendered":"

Documents<\/h1>\n

WivenLLM responds based on your documents. This page explains how to do this.
\nto enter, what happens to them, and how to achieve good results.<\/p>\n

\n

The two stages of importing<\/h2>\n

This is the point that most often causes confusion.<\/p>\n

Step 1 \u2014 Extraction.<\/strong> The file is converted to text. It joins the
\ndocument library<\/strong>, common to the instance. At this stage, it is not
\nqueryable in any space.<\/p>\n

Step 2 \u2014 Adding to a space.<\/strong> You select the document in the
\nAdd it to your library and workspace. It will then be split.,
\nvectorized and indexed for this space.<\/p>\n

The same document can be added to multiple spaces: it is only extracted once.
\ntimes, and indexed separately in each.<\/p>\n

Drag and drop directly into a conversation combines the two steps.
\nautomatically.<\/p>\n

\n

How to import<\/h2>\n

Files.<\/strong> Drag and drop into the conversation, or upload from the
\nDocument manager. Multiple files at once are accepted.<\/p>\n

Web pages.<\/strong> Paste a URL. The page content is extracted; ads and
\nNavigation menus are kept out of the way as much as possible.<\/p>\n

Complete site.<\/strong> A recursive import follows links from a starting page,
\nto a depth and number of pages that you specify. Use with
\nMeasurement: depth makes volume grow very quickly.<\/p>\n

External sources.<\/strong> Connectors import directly from:<\/p>\n\n\n\n\n\n\n\n\n\n\n
Source<\/th>\nWhat is imported<\/th>\n<\/tr>\n<\/thead>\n
Code repositories (GitHub, GitLab)<\/td>\nFiles in a repository, filterable by branch and by extension<\/td>\n<\/tr>\n
Confluence<\/td>\nPages of a space<\/td>\n<\/tr>\n
Company Wikis<\/td>\nPages and tree structure<\/td>\n<\/tr>\n
Document management (Paperless-ngx)<\/td>\nDocuments and metadata<\/td>\n<\/tr>\n
Note chests (Obsidian)<\/td>\nNotes and links between notes<\/td>\n<\/tr>\n
YouTube videos<\/td>\nTranscription<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n

From the API.<\/strong> Uploading a file, plain text, or URL, for
\nPower WivenLLM from your own applications.
\n\u2192 API developer<\/a><\/p>\n

\n

Accepted formats<\/h2>\n

Text and office automation<\/strong> \u2014 .txt<\/code>, .md<\/code>, .csv<\/code>, .json<\/code>, .html<\/code>, formats of
\ntechnical documentation, .pdf<\/code>, .docx<\/code>, .pptx<\/code>, .xlsx<\/code>, OpenDocument formats.<\/p>\n

Messaging and books<\/strong> \u2014 archives .mbox<\/code>, digital books .epub<\/code>.<\/p>\n

Images<\/strong> \u2014 .png<\/code>, .jpg<\/code>, .webp<\/code>, from which the text is extracted by
\ncharacter recognition.<\/p>\n

Audio and video<\/strong> \u2014 main audio formats and .mp4<\/code>, transcribed
\nautomatically. A meeting can also be recorded directly from
\nthe application \u2192 Meetings<\/a>.<\/p>\n

\u2192 Detailed list<\/a><\/p>\n

What works less well<\/h3>\n