Skip to main content
A workspace gives each portal user their own personal environment — a place to upload documents, build a personal knowledge base, run interfaces against their own data, and organize findings in tables. A credit analyst can upload loan documents and query them through the interfaces you configured. A researcher can build a personal knowledge base from papers and run analysis across them. The workspace makes this possible, all without leaving the portal.

​
Enabling the workspace

Workspace is off by default. To enable it, open the portal editor and turn on the Workspace toggle. Then click Configure to set it up.

​
Configuring the knowledge base

The knowledge base is the foundation of each user’s workspace. Click Configure to set it up.

​
Permanent settings

These are locked after the workspace is created — choose carefully:
  • Embedding Model (required): How documents are converted into searchable vectors. Defaults to voyageai/voyage-4-lite.
  • Advanced Document Analysis (beta): Enable deeper parsing for complex file types like scanned PDFs or mixed-format documents.
  • Hybrid Search: Combine vector and keyword search for more accurate retrieval.

​
Default settings

Users can adjust these later:
  • Chunk Size: How large each document chunk is when ingested. Defaults to 400.
  • Chunk Overlap: How much consecutive chunks overlap, to preserve context across chunk boundaries. Defaults to 0.
  • Splitter Method (required): How documents are split into chunks. Defaults to Markdown.
  • Processing Model (required): The model used during document processing. Defaults to Default.
  • Generate Chunk Level Metadata: Automatically generate metadata for each chunk during ingestion.
  • Max Rows Per Chunk: Set a limit on rows per chunk when processing tabular data.
  • Apify Key: Enter your Apify API key if you want to enable web scraping integrations.

​
Configuring interfaces

Choose which interfaces will be ready to use in each user’s workspace from day one. Click Add Interface to browse and select the interfaces you want to prepopulate. These interfaces will automatically query the user’s personal knowledge base — not the shared organization one.
This means each user’s results come from their own uploaded documents, keeping data isolated between users.

​
Table templates

Give users a starting structure for their tables instead of making them build from scratch. Click Add Template to include one of your existing tables as a template.

​
Enable table in workspace

Turn this on to give each user a personal table in their workspace — useful when users need to collect, organize, or analyze data alongside their knowledge base.

​
What the workspace looks like inside

Once a user creates their workspace, they get their own knowledge base, preconfigured interfaces, and tables — all in one place.
Workspaces keep data isolated between users. Each person’s uploaded documents, tables, and query results are private to their own workspace.