{ "trigger": "watch_files", "folder_id": "your_shared_drive_folder_id", "filter": "mimeType='application/pdf'" }
Once detected, pass the file to Chat With PDF by Copilot.us via their API. This tool reads the entire paper and returns structured sections: title, abstract, introduction, methodology, results, discussion, and conclusion.
POST https://api.chatpdf.com/v1/documents { "file": "base64_encoded_pdf_content", "metadata": { "filename": "paper_title.pdf" } }
Save the document ID returned by Chat With PDF; you'll reference it in the next step.
### Step 2: Extract and Simplify Complex Passages
Now you need human-readable explanations of the paper's core ideas. Use Explainpaper's API (or integrate manually through their interface) to highlight key passages. Since automation requires API access, you can alternatively send structured text through GPT-4.1 with a system prompt mimicking Explainpaper's function. Set up an n8n HTTP request node with Claude Opus 4.6 as your extraction backbone:
POST https://api.anthropic.com/v1/messages { "model": "claude-opus-4.6", "max_tokens": 1024, "messages": [ { "role": "user", "content": "Explain this academic passage in simple terms a first-year student can understand:\n\n[PASSAGE FROM PAPER]" } ] }
Claude will handle technical jargon beautifully and produce explanations ready for your lesson notes. Store these explanations in a structured format: passage ID, original text, simplified explanation.
### Step 3: Generate Exam Questions
With explanations in hand, generate multiple-choice and short-answer exam questions. Use GPT-4.1 mini to keep costs reasonable:
POST https://api.openai.com/v1/chat/completions { "model": "gpt-4.1-mini", "messages": [ { "role": "system", "content": "You are an expert educator. Generate 3 multiple-choice exam questions and 2 short-answer questions based on this academic content. Format as JSON with 'question', 'options' (for MC), 'correct_answer', and 'difficulty' (easy/medium/hard)." }, { "role": "user", "content": "[SIMPLIFIED PASSAGE FROM STEP 2]" } ] }
Parse the JSON response and store it in a questions database (a simple Google Sheet works, or use Supabase for more scale).
### Step 4: Create Flashcard Decks
Feed your explanations and extracted concepts into AnkiDecks AI to generate flashcard decks automatically. AnkiDecks accepts text input and outputs .apkg files (Anki package format). Set up an n8n node that batches your simplified explanations and sends them to AnkiDecks:
POST https://api.ankidecksai.com/generate { "format": "anki_package", "content": { "deck_name": "Paper_Title_Key_Concepts", "cards": [ { "front": "What is [key term]?", "back": "[definition from simplified explanation]", "source": "paper_title" } ] } }
### Step 5: Consolidate and Store
The final step pulls everything together. Create a summary document in Google Docs or Notion that includes: 1. Paper metadata (title, authors, link)
2. Simplified explanations of each section
3. Exam questions (in a downloadable format)
4. Link to the generated Anki deck
5. Tags for curriculum alignment Use n8n's Google Docs or Notion nodes to write this summary automatically.
{ "document": { "title": "Paper: [Title] - Teaching Materials", "body": [ { "type": "heading", "content": "Simplified Explanations" }, { "type": "paragraph", "content": "[Explanations from Step 2]" }, { "type": "heading", "content": "Exam Questions" }, { "type": "paragraph", "content": "[Questions from Step 3]" } ] } }
The entire flow runs on a schedule or webhook trigger. After uploading a paper, you'll have a complete teaching package within 10 minutes, ready to import into your LMS or print for students.
## The Manual Alternative
If you prefer more control over explanations and question quality, skip the automation and instead use these tools sequentially by hand. Start with Chat With PDF to skim the paper, use Explainpaper to clarify specific dense sections, and manually write your exam questions and flashcards. This takes longer but gives you the chance to align questions precisely with your learning objectives and add context from your other course materials. Alternatively, use QuillBot to paraphrase complex passages into simpler language before feeding them to AnkiDecks, which often produces better flashcards when the input text is already simplified.
## Pro Tips
### Rate limits and batch processing.
OpenAI and Anthropic rate-limit API calls.
If you're processing multiple papers, batch your requests or spread them across an hour using n8n's Delay node. AnkiDecks AI has a 10-call-per-minute limit on their standard tier; stay under it by queuing papers one at a time.
### Quality control on explanations.
Claude Opus 4.6 produces better explanations than GPT-4.1 mini, but costs more. Use Opus for introductions and key methodology sections, then use GPT-4.1 mini for results and discussion. Route by section type in n8n with conditional logic.
### Verify Anki deck imports.
Always download the generated .apkg file and open it in Anki Desktop before sharing with students. Occasionally, field mapping fails silently. A quick manual check prevents classroom chaos.
### Cost optimisation.
If processing many papers, switch to cheaper models where quality won't suffer. GPT-4.1 nano handles simple summarisation tasks that don't require deep reasoning. Gemini 2.5 Flash costs less than GPT-4.1 mini and handles question generation well. Test both before committing.
### Store explanations for reuse.
Save every Claude-generated explanation to a searchable database. If next semester you find a similar paper, you can retrieve and adapt an old explanation rather than regenerating from scratch, saving API calls.
## Cost Breakdown
| Tool | Plan Needed | Monthly Cost | Notes |
|------|-----------|--------------|-------|
| n8n | Free or Self-Hosted | £0–£50 | Free tier sufficient for 1-2 papers per week; paid tier if processing daily |
| Chat With PDF by Copilot.us | Pro | £15–30 | Supports concurrent file uploads and longer documents |
| Claude API (Anthropic) | Pay-as-you-go | £5–20 | Depends on explanation length; Opus is ~3p per 1K tokens |
| GPT-4.1 mini (OpenAI) | Pay-as-you-go | £3–10 | Cheaper than Opus; ideal for question generation |
| AnkiDecks AI | Standard | £10–15 | Covers unlimited deck generation up to 50 cards per request |
| QuillBot | Plus (optional) | £10 | Only if manually paraphrasing; skip if using Claude |
| Google Drive / Docs storage | Free or Workspace | £0–6 | Included with most Google Workspace plans |
| **Estimated Total** | | **£43–131** | Processing 4 papers monthly; adjust based on paper length | This workflow transforms the four-hour-per-paper manual process into a hands-off automation that frees you to focus on pedagogy rather than formatting. Start with one paper, validate the output quality, then expand.