🔬 AI-powered auto-generate exam questions and answers from Google Docs with Gemini

⚡ 3,432 views · 🔬 Document Extraction & Analysis

Description

This workflow automates the creation of exam questions (both open-ended and multiple-choice) from educational content stored in Google Docs, using AI-powered analysis and vector database retrieval

This workflow saves educators hours of manual work while ensuring high-quality, curriculum-aligned assessments. Let me know if you’d like help adapting it for specific subjects!

Use Cases

Educators: Rapidly generate quizzes, midterms, or flashcards.
E-learning platforms: Automate question banks for courses.
Corporate training: Create assessments for employee onboarding.

Technical Requirements:

APIs: Google Gemini, OpenAI, Qdrant, Google Workspace.
n8n Nodes: LangChain, Google Sheets/Docs, HTTP requests, code blocks.

This workflow combines AI efficiency with human-curated quality, making it a powerful tool for modern education and training.

Advantages of This Workflow

✅ Fully Automated Exam Generation: From document to fully formatted quiz content with no manual intervention.
✅ Supports Comprehension and Critical Thinking: Questions are designed to go beyond factual recall, including inference and application.
✅ Uses AI and RAG for Accuracy: Ensures that answers are grounded in the document content, reducing hallucination.
✅ Seamless Google Integration: Pulls content from Google Docs and writes outputs to Google Sheets.
✅ Scalable for Any Subject: Works with any article or content domain as input.
✅ Modular and Customizable: Can be easily adapted to generate different question types or to use other LLMs or storage systems.

How It Works

Document Ingestion:
- The workflow starts by fetching an educational document (e.g., textbook chapter, lecture notes) from Google Docs.
- Converts the document to Markdown for structured processing.
AI Processing:
- Splits text into chunks and generates vector embeddings (via OpenAI) for semantic analysis.
- Stores embeddings in Qdrant (vector database) for retrieval.
Question Generation:
- Open-ended questions: Google Gemini AI creates 10 critical-thinking questions.
- Multiple-choice questions: Generates 10 MCQs (1 correct + 3 plausible distractors) using RAG to validate answers against the vector DB.
Answer Validation:
- For open questions: Retrieves context-aware answers from the vector store.
- For MCQs: Ensures distractors are incorrect but believable via AI cross-checking.
Output:
- Saves questions/answers to Google Sheets in two tabs:
  - Open questions: Question + AI-generated answer.
  - Closed questions: MCQ + options + correct answer.

Set Up Steps

Prerequisites:
- APIs/Accounts:
  - Google Workspace (Docs + Sheets).
  - OpenAI (for embeddings).
  - Google Gemini (for question generation).
  - Qdrant (vector DB – self-hosted or cloud).
- n8n Nodes: Ensure LangChain, Google Sheets/Docs, and HTTP request nodes are installed.
Configure Connections:
- Link credentials for:
  - Google Docs/Sheets (OAuth2).
  - OpenAI (API key).
  - Google Gemini (API key).
  - Qdrant (URL + API key).
Customize Input:
- Replace the default Google Doc ID in the “Get Doc” node with your source document.
- Adjust chunk size/overlap (Token Splitter node) for optimal text processing.
Tweak Question Generation:
- Modify prompts in:
  - “Open questions” node: Adjust criteria (e.g., difficulty, question types).
  - “Closed questions” node: Edit MCQ formatting rules.
Output Settings:
- Update the Google Sheet ID in “Write open” and “Write closed” nodes.
- Map columns in Google Sheets to match question/answer formats.
Run & Automate:
- Trigger manually (“Test workflow”) or schedule periodic runs (e.g., for updated content).

Need help customizing?

Contact me for consulting and support or add me on Linkedin.

🔗 Nodes Used

Google Sheets, HTTP Request, Google Docs, AI Agent, Basic LLM Chain, Question and Answer Chain

📥 Import

Download workflow.json and import into n8n: Workflow menu → Import from File

📖 Importing guide · 🔑 Credential setup