AI Workflow for Document Summarization

Difficulty: Beginner · Setup time: 20 min · Steps: 4

Reports, contracts, and research papers pile up faster than anyone can read them. This workflow monitors a Google Drive folder — whenever a new PDF or document is uploaded, OpenAI extracts the key points, generates an executive summary, and saves it to Notion alongside the original file link. Your team gets the insights without reading every page.

What to Expect

  • 2 min — to extract key insights from a 50-page report
  • 100% — of uploaded documents summarized — nothing sits unread

Tools Used in This Workflow

How It Works — Step by Step

  1. Step 1: Detect new document in Google Drive

    Tool: Google Drive

    n8n monitors a specific Google Drive folder via polling. When a new file is uploaded (PDF, DOCX, or Google Doc), the file name, type, and content are extracted. Google Docs content is fetched directly; PDFs are converted to text using a parsing step.

  2. Step 2: Summarize with OpenAI

    Tool: OpenAI

    The document text is sent to OpenAI with a prompt requesting: a 3-sentence executive summary, 5 key points, any action items or decisions mentioned, and a recommended follow-up. For long documents, the text is chunked and summarized in passes.

  3. Step 3: Save summary to Notion

    Tool: Notion

    A new Notion page is created with the document title, upload date, Google Drive link, the full summary, key points as a bulleted list, and any extracted action items. The page is added to your team's knowledge base database for future searchability.

  4. Step 4: Share via Slack

    Tool: Slack

    A Slack message to the relevant channel includes the document title, the 3-sentence summary, a link to the Notion summary page, and a link to the original Google Drive file. Your team gets the gist in 10 seconds and can click through for more detail.

Pro Tips for This Workflow

  • Create separate Drive monitoring folders by document type (contracts/, reports/, research/) so the OpenAI prompt can be customized per type — contract summaries need different extraction logic than research reports.
  • Ask OpenAI to also output a 'relevance score' (1–5) for each document relative to your current strategic priorities — this helps prioritize which summaries to read when volume is high.
  • For very long documents, use n8n's split-in-batches node to chunk text into 3,000-word segments, summarize each chunk, then synthesize the chunk summaries into a final executive summary.

What You'll Need

  • Google Drive account — Document upload trigger
  • OpenAI account — Document summarization
  • Notion account — Summary storage
  • Slack account — Team notification
  • Basic automation experience, or a vetted expert to build it for you

Related Video Tutorial

Auto-Summarize Documents from Google Drive with OpenAI

Frequently Asked Questions

What file formats does this workflow support beyond PDF?

Google Docs, Google Sheets (as text), and .docx files can all be accessed via Google Drive API and parsed to text. For PDFs, use pdf-parse (in n8n's Code node) or a service like Adobe PDF Extract API for better accuracy on complex layouts. Plain .txt and .csv files are natively readable without any parsing.

How does n8n extract text content from a PDF in Google Drive?

n8n can download the file binary from Google Drive and pass it to OpenAI's file upload endpoint if it's under 20MB. Alternatively, use n8n's HTTP Request node to call Google Drive's 'Export' endpoint for Google Docs (converts to plain text automatically) or integrate pdf-parse via the n8n Code node for PDFs.

Can I summarize emails or Slack messages with this workflow?

Yes — for emails, monitor Gmail with n8n and trigger summarization on emails with specific labels or from specific senders. For Slack, use the Slack 'Read Channel Messages' node to pull the last 50 messages from a channel and summarize them as a daily digest posted to Notion or back to Slack.

How do I ensure sensitive document content isn't retained by OpenAI?

Enable OpenAI's zero data retention (ZDR) option available on their Enterprise and API tiers. With ZDR enabled, OpenAI does not retain your inputs or outputs for training. For highly sensitive documents (legal, financial, medical), consider using a self-hosted open-source LLM (Ollama + Llama 3) instead of the OpenAI API.

Want This Workflow Built for You?

Our vetted automation engineers can set up this exact workflow in days — customized to your tools and process. Free consultation, honest quote, fast delivery.

Get matched free — no commitment

More AI-Powered Workflows workflows | All workflow guides | Find automation experts