n8n Workflow: Scrape Paul Graham Essays & Summarize with Claude

Scrape the latest Paul Graham essays and turn them into Claude summaries—automatically

This n8n workflow uses Notte (real-browser scraping) to fetch the most recent Paul Graham essays and then produces clean, per-essay summaries with Claude Haiku. No brittle CSS selectors—works even on JavaScript-heavy pages.

What this workflow does

• Runs from n8n: Start by clicking Execute in n8n (or swap in a Schedule trigger to run daily).
• Fetch the latest essay links with Notte: The first Notte node opens paulgraham.com/articles.html and returns the 10 most recent essays as structured JSON using a short instruction + JSON schema (no CSS/HTML parsing).
• Prepare a demo-sized run: Split Out turns the list into one item per essay, and Limit keeps the first three to keep costs low.
• Fetch full essay text in a real browser: A second Notte node opens each URL and returns the full article text as markdown.
• Summarize with Claude: A summarization chain splits long text into chunks and summarizes each essay with Claude Haiku 4.5 (you can use any chat model).
• Output cleanup: Merge pairs each summary with its title and URL, and the workflow outputs one item per essay (title, url, summary).

Use cases

• Create a daily reading digest of Paul Graham essays with n8n automation.
• Test reliable scraping for JavaScript-heavy sites using Notte’s real-browser approach.
• Build a reusable pattern for scrape → summarize → store/display pipelines for SaaS teams.

Technical details

• Notte: Uses n8n-nodes-notte (community node) with Notte API credentials from console.notte.cc.
• Summarization: Anthropic integration via an Anthropic Chat Model node (Claude Haiku 4.5 referenced).
• n8n nodes: Set, Limit, Merge, Split Out, and a Sticky Note for documentation.