Overview
Content management is an essential step in configuring an AI coach. Content forms the knowledge base that the coach relies on for teaching.
Content Sources
- File upload — PDF, DOCX, PPT/PPTX, TXT (up to 2 GB per file).
- URL crawling — Add a URL with the option to crawl linked pages.
- Connections — Google Drive, Rise Up, Notion, OneDrive/SharePoint, Zoom Docs.
Connection Folder Tree
When you add content via a connection (Google Drive, Notion, OneDrive/SharePoint, Zoom Docs, etc.), files are displayed as an interactive folder tree that mirrors the folder structure of the source.
- Clickable folders — Click a folder to expand or collapse it and navigate the hierarchy.
- Dynamic loading — When opening a connection, a loading animation indicates that the structure is being fetched from the source.
- Load error — If the structure cannot be retrieved (e.g., access was revoked), an error message appears with a retry button.
Subscribed Folders (automatic sync)
When you check a folder in the tree, the coach subscribes to that folder and all of its subfolders — not just to the files present at selection time.
- New files picked up — Files added later to a subscribed folder are indexed at the connection's next sync, and the coach is retrained automatically.
- Mixed selection — You can combine subscribed folders with individual files selected outside those folders.
- Limit — A coach can follow at most 100 effective files (selected files + contents of subscribed folders). Selections beyond this cap are rejected.
Content Status
- Not indexed (gray) — Not yet processed.
- Ready (green) — Indexed and available.
- Processing (spinning) — Indexing in progress.
- Error (red) — Problem during indexing. The cause is now shown directly on the affected content: for example site blocking automated access, link not found (404), unreadable media, or no text detected. The technical detail is available on hover. Fix the source (replace the URL, check the file) and then relaunch the sync.
Check the Extracted Text
For each processed content item (uploaded file, URL or item from a connection), an eye icon (tooltip "View text") appears in the coach's knowledge base content list.
When to use it: to check what the coach actually "reads" in a document (for example a scanned PDF or a dynamically loaded web page), or to understand why a coach answer is incomplete.
Prerequisites: edit access to the coach, and the content must have been processed by Forge.
- Extracted text — Click the icon to open a window showing the text extracted by Forge, split into segments and paginated (e.g. "Segments 1-10 of 42").
- Crawled pages — For a URL added with the crawl option, the window shows the number of crawled pages and lists them.
- No text — The message "No extracted text for this content yet" means extraction produced nothing (or is not finished). If the status is Error, check the cause shown above and fix the source.