NotebookLM API · unofficial, by Cotlet
Create your own NotebookLM API
NotebookLM: the only official API is for NotebookLM Enterprise, in preview; none was found for a personal Google account (checked 20 Nov 2026). With Cotlet you can build your own. Do a job once with your AI agent in your NotebookLM account, save it as an endpoint, and call it from n8n, Make, Zapier or your own code. Each endpoint runs in your own signed-in NotebookLM session and returns only what your login can see.
Free 7-day trial, 2,000 credits, no card
Cotlet is an independent service. It is not affiliated with, endorsed by or sponsored by NotebookLM. NotebookLM is a trademark of its owner. If NotebookLM's own integrations cover what you need, use them first.
curl -X POST "https://cotlet.dev/api/v1/custom/notebook-sources" \
-H "Authorization: Bearer $COTLET_API_KEY" \
-H "Content-Type: application/json" \
-d '{"notebook": "Thesis"}' const url = "https://cotlet.dev/api/v1/custom/notebook-sources";
const res = await fetch(url, {
method: "POST",
headers: {
Authorization: `Bearer ${process.env.COTLET_API_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({ notebook: "Thesis" }),
});
const { data } = await res.json(); import os
import requests
key = os.environ["COTLET_API_KEY"]
res = requests.post(
"https://cotlet.dev/api/v1/custom/notebook-sources",
headers={"Authorization": f"Bearer {key}"},
json={"notebook": "Thesis"},
)
data = res.json()["data"] {
"data": [
{
"title": "Sleep and memory, 2024",
"type": "PDF",
"added": "2026-11-18"
},
{
"title": "Chapter 2 notes",
"type": "note",
"added": "2026-11-19"
}
]
} 01 Build
How do you build a NotebookLM endpoint with Cotlet?
-
Sign in to NotebookLM in your Cloud Browser.
You type your password and NotebookLM's 2-Step Verification code yourself. Cotlet keeps the signed-in session, encrypted, never the password.
-
Do the job once with your agent.
For example: "List the sources in my Thesis notebook with their type and the date I added them."
-
Save it as an endpoint.
Give it a name, choose its inputs, such as a notebook name or a question, and the fields it returns.
-
Call it with your Cotlet API key.
Each endpoint gets its own address,
POST https://. You send JSON in and get JSON back, from the same signed-in NotebookLM session.cotlet.dev/ api/ v1/ custom/ <name>
02 Endpoints
Example endpoints you could build for NotebookLM
Cotlet ships no fixed list of NotebookLM endpoints. You decide which endpoints you build and what each one returns.
-
POST
/api/v1/custom/notebook-sources Sources in a notebook
Title, type and date added for each source
-
POST
/api/v1/custom/ask-notebook Ask a notebook
The answer from your sources, with the sources it cites
-
POST
/api/v1/custom/list-notebooks Your notebooks
Each notebook with its number of sources
-
POST
/api/v1/custom/studio-outputs Studio outputs
Audio overviews, reports and tables in a notebook
-
POST
/api/v1/custom/writesadd-source Add a source
A web page or YouTube link added to the notebook you name
A call to an endpoint you built could look like this:
curl -X POST "https://cotlet.dev/api/v1/custom/notebook-sources" \
-H "Authorization: Bearer $COTLET_API_KEY" \
-H "Content-Type: application/json" \
-d '{"notebook": "Thesis"}' The answer is JSON, with your fields under data:
{
"data": [
{
"title": "Sleep and memory, 2024",
"type": "PDF",
"added": "2026-11-18"
},
{
"title": "Chapter 2 notes",
"type": "note",
"added": "2026-11-19"
}
]
} When you save an endpoint, Cotlet gives you its curl example and an OpenAPI description to paste into n8n, Make or your code.
03 Uses
What can you use your NotebookLM endpoints for?
- Friday Cotlet endpoint notebook
A reading list that files itself.
New links from your reading list added to a notebook every Friday.
- Notes app endpoint cited answer
Answers in your notes app.
A question sent to a notebook, the cited answer saved to your notes.
- Weekly endpoint Google Sheets
A sheet of your sources.
Every source in a notebook listed in Google Sheets for your bibliography.
- Chat agent endpoint notebook
Your own study assistant.
A chat agent that asks your notebook before it answers you.
04 Access
What can an endpoint see and change?
An endpoint works in your own NotebookLM session, so it can read and change only what your NotebookLM login can. You choose what each endpoint does when you build it.
An endpoint that changes something, such as adding a source, first runs as a dry run: Cotlet shows you each request it would send, and nothing is sent. You allow the writes once in the Cotlet console and set a daily cap. A call over the cap fails whole, never half done.
05 Official API
Does NotebookLM have an official API?
NotebookLM: Google documents preview notebook methods for NotebookLM Enterprise that need a Gemini Notebook Enterprise license; none was found for a personal Google account (checked 20 Nov 2026). Reports export to Google Docs. The full picture with sources: Does NotebookLM have an API or MCP server?
Want ChatGPT or Claude to work in NotebookLM instead of code? See the NotebookLM MCP server.
06 Pricing
What does it cost?
- per successful call
- 10 credits
- when a call fails
- 0
- in the smallest pack
- ≈ 500 calls
- free trial, no card
- 7 days
A call to your endpoint costs 10 credits when it succeeds and nothing when it fails, with no browser minutes on top. Credits come in one-time packs from $29 for 5,000 credits, so a pack covers about 500 successful calls. The free trial gives you 2,000 credits for 7 days, no card.
NotebookLM API with Cotlet: questions
Is this an official NotebookLM API?
No. Cotlet is independent and not affiliated with NotebookLM. The endpoints you build run in your own signed-in NotebookLM account.
Do I need to code?
Not to build an endpoint: you do the job once with your agent and save it. To call it, you need a tool that sends HTTP requests, such as n8n, Make, Zapier or a script.
What happens when NotebookLM changes its screens?
An endpoint works while NotebookLM works the way it did when you built it. If NotebookLM changes and a call fails, you get an email. After 3 failures in a row the endpoint turns off until you rebuild it with your agent or in the console. A failed call costs nothing.
How long can a call take?
Up to 90 seconds, and the answer can be up to 1 MB of JSON.
Build your first NotebookLM endpoint
Sign in once, do the job with your agent and save it as an endpoint.
Try it free nowNo card. Takes about 5 minutes.