I replaced a $49/month PDF parsing API with 10 lines of code in n8n to extract text from PDFs.
Most people stick to prebuilt nodes instead of writing logic.
When something doesn’t fit, they just keep adding nodes until it works.
At some point the workflow has 12 nodes doing what 10 lines of code would have handled.
The Code node lets you do exactly what you need.
You don’t even have to write it yourself. ChatGPT, Claude, or any LLM can generate it in seconds.
This was about 10 lines using a small npm package.