Guide: n8n for Auto-Vectorizing PDFs from Google Drive to Zyrabit RAG

Integrate n8n (open source workflow) into your Oracle Zyrabit node to: monitor Drive folder → extract PDF text → embeddings → insert into local ChromaDB (Zyrabit RAG). Everything sovereign, without uploading data to external clouds. Runs on the same Forge docker-compose (observability profile).

Objective: Automates indexing of contracts/notes → Zyrabit chat answers "Does this PDF violate CNBV?" with automatic citations. Time: 20 min setup.

Phase 1: Install n8n on Forge Stack

Extend docker-compose.yml: Add n8n service to observability profile.

Phase 2: Google Drive Credentials

Setup Google Cloud Console Service Account and share "Zyrabit-Ingesta" folder.

Phase 3: n8n Workflow "PDF → Automatic RAG"

Import provided JSON workflow in n8n UI.

n8n Steps:

1. Credentials: New > Google Drive OAuth2 > use service account JSON.
2. Google Drive Trigger: Watch folder ID.
3. Code Node: Install pdf-parse in n8n container.
4. HTTP Nodes: Point to internal Forge endpoints (LiteLLM/Chroma).
5. Save & Activate.

Phase 4: Validation and Improvements

End-to-end Test: Upload contract PDF to "Zyrabit-Ingesta" Drive. n8n log: Generated embeddings → Chroma upsert. Zyrabit Chat: "What does the CNBV PDF say?" → automatic RAG citation.

Result: Drive → Auto-indexed PDF → RAG ready for sovereign chat. Ideal for PoCs.