Does NotebookLM Have an API? 2026 Status

The current answer: three different API surfaces

The important distinction is access, not whether any Google endpoint exists. Google's current developer documentation places notebook APIs under Gemini Notebook Enterprise, and marks audio-overview creation as a Pre-GA Preview feature. The standalone podcast endpoint is a different API: Google says it does not require a Gemini Notebook Enterprise license, but it is deprecated and closed to new allowlisting.

Where AutoContent API fits

AutoContent API's NotebookLM-style API is a separate product, not a wrapper around the consumer NotebookLM interface. It accepts source material through a documented REST endpoint and can return podcasts, transcripts, summaries, quizzes, study assets, videos, infographics, and slide decks. Product capabilities and pricing belong to AutoContent API; Google's terms and availability apply only to Google's own APIs.

Authentication: create a key and keep it server-side

Create an AutoContent API key in the AutoContent API app, store it as a server-side secret, and send it as a Bearer token. Do not expose the key in browser JavaScript, public source control, analytics, or URLs.

Request and response example

This request starts a two-host podcast from a webpage and includes callback metadata your own application can use to reconcile the job:

curl -X POST "https://api.autocontentapi.com/content/Create" \
  -H "Authorization: Bearer $AUTOCONTENT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "outputType": "audio",
    "resources": [
      { "type": "website", "content": "https://example.com/your-article" }
    ],
    "text": "Create a concise two-host podcast about the source.",
    "callbackData": "workspace=docs;job=article-123"
  }'

Creation is asynchronous. Save the returned request ID immediately:

{
  "request_id": "550e8400-e29b-41d4-a716-446655440000"
}

Status polling and webhook lifecycle

1. Send the create request and persist its request_id.
2. Poll GET /content/Status/{request_id} at a reasonable interval.
3. Treat status: 100 as complete and store the returned asset URLs.
4. Stop on an error response or your own timeout instead of polling forever.
5. Register a webhook for prompt delivery, while keeping polling as reconciliation for missed callbacks.

curl -X GET \
  "https://api.autocontentapi.com/content/Status/550e8400-e29b-41d4-a716-446655440000" \
  -H "Authorization: Bearer $AUTOCONTENT_API_KEY"

{
  "id": "550e8400-e29b-41d4-a716-446655440000",
  "status": 100,
  "audio_url": "https://data.autocontentapi.com/example.mp3"
}

Register the callback URL once. Your receiver should acknowledge quickly, deduplicate by request ID, and queue any expensive downstream work:

curl -X POST "https://api.autocontentapi.com/content/Webhook" \
  -H "Authorization: Bearer $AUTOCONTENT_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "url": "https://example.com/api/autocontent/webhook" }'

For retries, callback handling, and production reconciliation, use the detailed webhooks, polling, and status guide.

Supported inputs and outputs

AutoContent API can ingest sources such as webpages, PDFs, documents, YouTube videos, feeds, and raw text. The available output depends on the request and plan, and includes:

• Multi-host audio podcasts and transcripts
• Summaries, FAQs, study guides, timelines, and briefing documents
• Videos, infographics, slide decks, quizzes, and structured text outputs
• Shareable asset URLs and status metadata for downstream workflows

Limitations to plan for

• AutoContent API is independent from Google. It does not provide access to a user's consumer NotebookLM notebooks or reproduce Google's implementation exactly.
• Google labels the Gemini Notebook Enterprise audio-overview API as Preview, and Google can change preview availability or behavior.
• The standalone Google Podcast API is deprecated and unavailable to new customers, even though its reference remains online.
• Generation is asynchronous. Completion time varies with source size, source accessibility, output type, and queue conditions.
• Restricted URLs, malformed files, or unsupported source content can fail ingestion. Use status errors and timeouts in your integration.
• AI-generated material can contain mistakes. Review factual, legal, medical, financial, or brand-sensitive output before publishing.

NotebookLM API FAQ