Build a PDF Document RAG System with Mistral OCR, Qdrant and Gemini AI
Knowledge & search workflow for n8n with 29 active steps. Starts with Manual Trigger, Chat Trigger, Execute Workflow Trigger. Uses Mistral AI, OpenAI, Google Gemini. Integrations include Mistral AI, OpenAI, Google Gemini, Qdrant, Google Drive.
What it does
29 active nodes · Starts with Manual Trigger, Chat Trigger, Execute Workflow Trigger.
Steps in the template include:
- Mistral Upload (HTTP Request)
- Mistral Signed URL (HTTP Request)
- Mistral DOC OCR (HTTP Request)
- When clicking ‘Test workflow’ (Manual Trigger)
Nodes and integrations
- HTTP Request ×5
- Manual Trigger
- Loop Over Items (Split in Batches) ×2
- Embeddings OpenAI ×2
- Default Data Loader
- Token Splitter
- Chat Trigger
- Question and Answer Chain
- Google Gemini Chat Model ×2
- Vector Store Retriever
- Qdrant Vector Store ×2
- Code
- Wait
- Execute Sub-workflow
- Execute Workflow Trigger
- Edit Fields (Set) ×3
- Summarization Chain
- Google Drive ×2
See the author’s step names
- Mistral Upload (HTTP Request)
- Mistral Signed URL (HTTP Request)
- Mistral DOC OCR (HTTP Request)
- When clicking ‘Test workflow’ (Manual Trigger)
- Loop Over Items (Loop Over Items (Split in Batches))
- Refresh collection (HTTP Request)
- Embeddings OpenAI (Embeddings OpenAI)
- Default Data Loader (Default Data Loader)
- Token Splitter (Token Splitter)
- When chat message received (Chat Trigger)
- Question and Answer Chain (Question and Answer Chain)
- Google Gemini Chat Model (Google Gemini Chat Model)
- Vector Store Retriever (Vector Store Retriever)
- Qdrant Vector Store1 (Qdrant Vector Store)
- Embeddings OpenAI1 (Embeddings OpenAI)
- Code (Code)
- Wait (Wait)
- Qdrant Vector Store (Qdrant Vector Store)
- Loop Over Items1 (Loop Over Items (Split in Batches))
- Execute Workflow (Execute Sub-workflow)
- When Executed by Another Workflow (Execute Workflow Trigger)
- Edit Fields1 (Edit Fields (Set))
- Create collection (HTTP Request)
- Summarization Chain (Summarization Chain)
- Google Gemini Chat Model1 (Google Gemini Chat Model)
- Set page (Edit Fields (Set))
- Set summary (Edit Fields (Set))
- Search PDFs (Google Drive) — fileFolder
- Get PDF (Google Drive) — download
Accounts and credentials
- Mistral AI access
- Mistral AI
- OpenAI access
- Google Gemini access
- Qdrant
- Google Drive
Credential types in the graph: mistralCloudApi, httpHeaderAuth, openAiApi, googlePalmApi, qdrantApi, googleDriveOAuth2Api
Check the HTTP request steps for any additional service accounts or API keys.
AI models and swaps
Mistral AI · Model or API operation
Model ID not specified in the template in Mistral Upload. Provider-specific step. A different provider requires edits and testing.
Mistral AI · Model or API operation
Model ID not specified in the template in Mistral DOC OCR. Provider-specific step. A different provider requires edits and testing.
OpenAI · Embeddings
Model ID not specified in the template in Embeddings OpenAI. Changing the embedding model may require rebuilding the vector index and matching its dimensions.
Google Gemini · Chat
models/gemini-1.5-flash in Google Gemini Chat Model. Potentially swappable with a compatible n8n chat model node. Test the replacement with the next step.
OpenAI · Embeddings
Model ID not specified in the template in Embeddings OpenAI1. Changing the embedding model may require rebuilding the vector index and matching its dimensions.
Google Gemini · Chat
models/gemini-2.0-flash-exp in Google Gemini Chat Model1. Potentially swappable with a compatible n8n chat model node. Test the replacement with the next step.
Resource information
- Credit basis
- Author listed by n8n
- Pricing
- Template access shown on n8n
- License
- Not specified
- Last record update
- Nov 5, 2026