Skip to main content

​
Overview

Collections let you group related documents into named sets. One document can belong to multiple collections — perfect for organizing by project, client, quarter, or any category that fits your workflow.
Many-to-many grouping. A single annual report can live in both “ACME Corp” and “Q4 2025 Filings” at the same time.

​
What You Can Do

Organize by Project

Group all documents for a deal, audit, or research project

Chat with Collections

Open a chat session scoped to a collection of documents

Flexible Membership

Add or remove documents at any time — no rigid folder hierarchy

CLI & API

Manage collections programmatically via the REST API or CLI

​
How It Works

  1. Create a collection with a name and optional description
  2. Add documents by their job IDs — existing completed extractions
  3. Query the collection to see all documents and their status
  4. Chat with the collection via CLI (okra chat -c "My Collection")

​
Example: Financial Analysis Pipeline

# Create a collection for Q4 earnings
okra collections create "Q4 2025 Earnings" -d "Quarterly filings for portfolio companies"

# Add completed extraction jobs
okra collections add "Q4 2025 Earnings" ocr-abc123 ocr-def456 ocr-ghi789

# List documents in the collection
okra collections show "Q4 2025 Earnings"

# Chat with the collection
okra chat -c "Q4 2025 Earnings" -m "Compare revenue growth across all companies"

​
Next Steps

CLI Reference

Full CLI command reference for collections

API Reference

Manage collections via the REST API