How to Turn Web Pages, YouTube Videos and RSS Feeds into NotebookLM Podcasts with n8n

October 4, 2026

Table of contents

  1. Introduction
  2. The two workflows
  3. How they work
  4. Set it up
  5. When something goes wrong
  6. What it costs
  7. More ways to use the NotebookLM API
  8. Frequently asked questions

Introduction

Two free n8n workflows send your links to NotebookLM through the useapi.net NotebookLM API (Gemini Notebook API) and bring back the finished podcast as an .m4a file. One is a form where you paste web pages and YouTube videos. The other watches a blog’s RSS feed and turns every week’s new posts into one episode. Both use only core n8n nodes and run on your own Google account, including a free one.

The episode below came from the RSS workflow, pointed at Google’s AI blog feed with format set to brief and length to short. Run once by hand, it picked up the two posts from the past week and produced a 1 min 43 s episode in 3 min 23 s.

“Google AI Fights Cyberattacks and Disease”, made by the RSS workflow from the Google AI blog feed, brief format, short length, 1 min 43 s, generated in 3 min 23 s

The two workflows

Workflow What it does File
Links to podcast A form: paste web pages and YouTube links, pick a format and a length, get the .m4a back notebooklm-podcast.json
RSS feed to weekly podcast Every Monday at 7:00, the newest posts of a feed become one episode notebooklm-rss-podcast.json

Both files are in the notebooklm-api repository on GitHub, MIT-licensed. They were tested on self-hosted n8n 2.41.

How they work

The form workflow asks for a title, the links (up to 50, one per line), the podcast format and the length. Then it:

  1. creates a notebook with POST /notebooks and adds the links with POST /sources.
  2. checks the notebook every 10 seconds until Google has read every source.
  3. starts the Audio Overview with POST /artifacts and polls GET /jobs every 20 seconds.
  4. downloads the .m4a as binary data for the next node.

The RSS workflow does the same from a Schedule Trigger. It reads the feed, keeps the posts published in the last lookbackDays days (7 by default), newest first and at most maxPosts (5), and puts them in one notebook named <showName>, <date>. A week with no new posts makes no episode. The schedule runs in your n8n instance’s time zone.

The formats are deep_dive, brief, critique and debate, and the lengths are short, default and long. The Settings node (form) or Feed settings node (RSS) also holds language, for example es, pt_BR or zh_Hans, and email, which you leave empty unless you want one specific Google account.

Set it up

  1. Get a useapi.net API token and connect your Google account with the NotebookLM setup.
  2. In n8n, import the file with Workflows → Import from File.
  3. Create a Header Auth credential, for example called useapi.net API token, with Name Authorization and Value Bearer <your API token>.
  4. Select that credential on the six HTTP Request nodes: Create notebook, Add sources, Check sources, Generate Audio Overview, Check job and Download audio.
  5. For the form, click Execute workflow to test it, or publish (activate) the workflow and open its form URL. For RSS, put your feed in the Feed settings node, click Execute workflow to try it, then publish (activate) it.

A short brief takes about 3 to 10 minutes. Add a Google Drive, S3, Telegram or podcast-host node after Download audio to send each episode where you need it. If you change the RSS schedule, change lookbackDays to match, for example a daily run with lookbackDays: 1.

When something goes wrong

  • A 401 on the first HTTP node means the credential is missing or the token is wrong. Check that the value starts with Bearer .
  • A 429 on Generate Audio Overview means the Google account is out of NotebookLM usage for now, and the run stops there. The response carries retryAt when Google names the reset time. To make more episodes, connect more Google accounts through the NotebookLM setup and leave email empty. New notebooks then go to the least busy account, so the next run can land on one that still has usage.
  • A failed job stops the workflow with the error Google returned.
  • A source that ends in error, for example a YouTube video without captions, is skipped and the episode uses the rest.
  • The workflows keep polling until the sources are read and the job finishes. Stop a run that has been polling for more than about 20 minutes.

What it costs

A flat $15/month for useapi.net, which covers every useapi.net API. There is no charge per episode. NotebookLM’s usage limits come from your Google plan, and each subscription lets you connect 3 Google accounts, up to 100 in total, to make more. See Run NotebookLM at scale.

More ways to use the NotebookLM API

  • A real example: one podcast, a video overview and an infographic from the same two sources.
  • The full tutorial: every Studio output, chat with citations and Deep Research, with curl.
  • GitHub examples: Node.js, Python, a Colab notebook and an agent skill for Claude Code and Codex.

Frequently asked questions

Does NotebookLM have an API?
Google offers a NotebookLM Enterprise API to Google Cloud customers with Enterprise licenses. Of the Studio outputs, it makes Audio Overviews only. The useapi.net NotebookLM API drives your own Google account instead, free or paid, and covers every Studio output.

Do I need n8n Cloud?
No. Both workflows use only core nodes. They were tested on self-hosted n8n and need nothing Cloud-only.

Start using the API

A single $15/month subscription covers every API on useapi.net — one token, all services. You connect up to 3 of your own provider accounts per service, and useapi.net handles the REST API, load balancing, and polling.

Subscribe — $15/month

Pricing, crypto payment, and volume options — full refund within the first 14 days if you have fewer than 50 successful generations.