Academy
3. Adding Documents: Upload and MCP Import
Learn to upload files via the drop zone, monitor processing progress, and alternatively import content from a connected integration with auto-sync.
When you open a knowledge base, you'll see a drop zone where you can add documents. Simply drag and drop your files into this area. Processing begins automatically in the background, and a progress bar lets you track each document's status. A document becomes searchable only once its status shows as processed. If a file is rejected on drop, it means the file type isn't allowed — your administrator configures which formats are accepted (typically PDF, Word, plain text, and Markdown).
Don't wait for one batch to finish before dropping more files. Processing runs in the background, so you can continue adding documents or even leave the page and come back later to check progress.
If a document fails to process, the issue is usually related to its format or size. Your first step should be to re-upload a cleaner version of the file (for example, re-export a PDF without scanned images, or simplify complex formatting). If multiple documents are stuck in a failed state, contact your administrator — they can trigger a reprocess from the admin panel.
Instead of uploading files manually, you can fill a knowledge base from a connected integration using MCP import mode. This option is available at creation time: switch to MCP import mode, select the connected service you want to pull content from, and the base will be populated automatically. The key advantage is auto-sync — the knowledge base stays up to date as the source content changes, without any manual re-upload on your part.
Choose MCP import when your source content lives in a tool your team already updates regularly (e.g., a wiki or project management tool). This way your knowledge base always reflects the latest version without anyone remembering to re-upload files.
Keep in mind that indexing documents consumes embedding credits proportional to the size of the content. Your plan and administrator settings also cap the number of documents per knowledge base. Before bulk-uploading, check your available credits and limits to avoid surprises. Searching the base afterward is lightweight and costs very little by comparison.
Open one of your knowledge bases (or create a new one). Drop a PDF or text file into the upload zone and watch the progress bar. Once the status shows "processed", use the search field on the base page to type a question and verify the document's content is retrieved correctly.
You now know two ways to populate a knowledge base: drag-and-drop upload with background processing and progress tracking, and MCP import with automatic sync from a connected service. Failed documents can be re-uploaded or reprocessed by an admin. Remember that indexing uses embedding credits proportional to content size, and always test retrieval with the built-in search field after adding documents.