Documents

The Documents section is your team's shared AI knowledge base. Upload files once and they become available as retrieval context across Smart Chat, Custom Agents, Workflows, Deep Research, and Bulk Processing.

What Documents does

When you upload a file to Documents, Hubrix:

1. Extracts the text content from the file
2. Chunks the text into segments of roughly 500 tokens
3. Embeds each chunk as a vector using a text embedding model
4. Stores the vectors in a searchable index

This process is called indexing. Once a document is indexed, any feature that supports RAG can query it — meaning the AI receives only the most relevant passages from your document rather than the whole file.

Uploading documents

1. Click Documents in the left sidebar.
2. Click Upload or drag and drop one or more files onto the upload zone.
3. Supported formats appear in the upload zone. Files outside these formats are rejected with an error message.
4. Watch the status column — new uploads start as Processing, then move to Indexed when ready.

Document sources

Documents in your library can come from two places:

• Manual uploads — files you drag in directly
• Connector syncs — files automatically synced from Google Drive or OneDrive

Connector-synced documents display a source badge (e.g., "From Google Drive") so you can tell where each file originated at a glance.

Using documents elsewhere in Hubrix

What's next

• Supported formats — which file types Hubrix accepts
• RAG explained — how retrieval-augmented generation works under the hood
• Connector-synced docs — documents sourced from Google Drive and OneDrive