Advanced

Document Workspace

A centralized file vault protected by Row Level Security. Powered by Large Context Window models that read the entire document at once.

Upload a meeting recording, PDF, video, table or Excel. Workspace transcribes speech, extracts key points, finds tasks, compares data and creates a summary you can use in Chat or Council.

Tip

For long materials, first create a clean Workspace summary and then send it to the Council. You get more accurate answers with less noise.

RAG vs. Large Context Window

This is the key distinction that determines how the AI interacts with your documents:

RAG vs. Large Context Window

RAG: Chat memory (RAG): documents are split into chunks, converted to embeddings, and only relevant snippets are pulled back into context. Ideal for a large archive across projects, but retrieval can miss context that isn't semantically close to the query.

Large Context: Document Workspace (Large Context Window): models like Gemini 1.5 Pro load the entire document at once — even hundreds of pages. The AI sees the complete structure, cross-references, numerical sequences, and edge details. No chunking, no lost context.

Rule of thumb: one long file → Workspace. Hundreds of files → RAG chat memory.

Row Level Security

Every file is protected in the database by RLS policies — you only see your own documents, even if the underlying vault is shared with your team. Encryption in transit and at rest is standard, and data stays in the EU.

Typical inputs

Audio and video from meetings — transcript, decisions, tasks.
PDFs and documents — summaries, fact extraction, questions over content.
Excel and tables — data checks, trends, anomalies and chart suggestions.
Outputs for Chat, Council, presentation or client portal.