notebooklm generate — create AI content

The generate command group creates AI content from your notebook sources. Most subcommands are asynchronous by default: they start generation and return immediately with a task_id, so you can continue working while the content is produced in the background. The only synchronous exception is mind-map, which completes instantly.

Generation tasks can take anywhere from seconds (mind-map) to 30+ minutes (complex video or audio). For LLM agents, avoid --wait and instead poll with artifact wait <id> in a separate step.

notebooklm generate <type> [description] [OPTIONS]

All generate subcommands support:

-s, --source ID — limit generation to specific sources (repeatable)
--json — return {"task_id": "...", "status": "..."} for scripting
--language CODE — override the output language (defaults to account setting or en)
--retry N — automatically retry on rate limits with exponential backoff
--wait — wait inline for completion (not recommended for agents)

audio — generate a podcast

Generate an audio overview in podcast format from your notebook sources.

notebooklm generate audio [description] [OPTIONS]

Option	Values	Default	Description
`--format`	`deep-dive`, `brief`, `critique`, `debate`	`deep-dive`	Podcast discussion format
`--length`	`short`, `default`, `long`	`default`	Approximate duration
`--language`	BCP-47 code	account setting	Output language
`-s, --source`	Source ID	all sources	Limit to specific source(s)
`--wait`	—	off	Wait for completion inline
`--json`	—	off	Machine-readable output
`--retry N`	integer	`0`	Retries on rate limit with backoff

# Basic podcast (async, returns immediately)
notebooklm generate audio

# Debate format with a custom instruction
notebooklm generate audio "Compare the two main viewpoints" --format debate

# Short podcast, wait for completion
notebooklm generate audio "Focus on key takeaways" --length short --wait

# Generate from specific sources only
notebooklm generate audio -s src_abc -s src_def

# Machine-readable output for scripting
notebooklm generate audio --json
# Output: {"task_id": "abc123...", "status": "pending"}

# Retry up to 3 times on rate limits
notebooklm generate audio --retry 3 --json

video — generate a video overview

Generate a video overview with animated visuals from your notebook sources.

notebooklm generate video [description] [OPTIONS]

Option	Values	Default	Description
`--format`	`explainer`, `brief`	`explainer`	Video format
`--style`	`auto`, `classic`, `whiteboard`, `kawaii`, `anime`, `watercolor`, `retro-print`, `heritage`, `paper-craft`	`auto`	Visual animation style
`--language`	BCP-47 code	account setting	Output language
`-s, --source`	Source ID	all sources	Limit to specific source(s)
`--wait`	—	off	Wait for completion inline
`--json`	—	off	Machine-readable output

# Kid-friendly explainer with kawaii style
notebooklm generate video "Explain for 5 year olds" --style kawaii

# Professional classic style, wait for completion
notebooklm generate video --style classic --wait

# Generate from specific sources
notebooklm generate video -s src_123 -s src_456

# Machine-readable output
notebooklm generate video --json

cinematic-video — generate a cinematic video

Alias for generate video --format cinematic. Generates a cinematic-style video overview. Accepts the same --style and other options as video.

notebooklm generate cinematic-video [description] [OPTIONS]

# Documentary-style cinematic video
notebooklm generate cinematic-video "Documentary about quantum physics"

# With a specific visual style
notebooklm generate cinematic-video "History of AI" --style watercolor --wait

slide-deck — generate a slide presentation

Generate a slide deck from your notebook sources. Download as PDF (default) or editable PPTX with notebooklm download slide-deck --format pptx.

notebooklm generate slide-deck [description] [OPTIONS]

Option	Values	Default	Description
`--format`	`detailed`, `presenter`	`detailed`	Slide template
`--length`	`default`, `short`	`default`	Approximate slide count
`--wait`	—	off	Wait for completion inline
`--json`	—	off	Machine-readable output

# Default detailed slide deck
notebooklm generate slide-deck

# Presenter slides with speaker notes
notebooklm generate slide-deck "Key findings for the board" --format presenter --wait

# Short slide deck as JSON
notebooklm generate slide-deck --length short --json

revise-slide — revise an individual slide

Revise a specific slide in an existing slide deck using a natural-language description. The slide deck must already be fully generated before you can revise individual slides.

notebooklm generate revise-slide <description> --artifact <id> --slide N [OPTIONS]

Option	Values	Description
`-a, --artifact ID`	Artifact ID	Required. The slide deck to revise
`--slide N`	Integer	Required. Zero-based slide index (0 = first slide)
`--wait`	—	Wait for revision to complete
`--json`	—	Machine-readable output

# Revise the first slide
notebooklm generate revise-slide "Move the title up" --artifact art123 --slide 0

# Revise the fourth slide and wait
notebooklm generate revise-slide "Remove the taxonomy table" --artifact art123 --slide 3 --wait

# Revise and get JSON output
notebooklm generate revise-slide "Simplify bullet points" --artifact art123 --slide 1 --json

Use notebooklm artifact list --type slide-deck to find the artifact ID for an existing slide deck.

quiz — generate a quiz

Generate a multiple-choice quiz based on your notebook sources.

notebooklm generate quiz [description] [OPTIONS]

Option	Values	Default	Description
`--difficulty`	`easy`, `medium`, `hard`	`medium`	Question difficulty
`--quantity`	`fewer`, `standard`, `more`	`standard`	Number of questions
`-s, --source`	Source ID	all sources	Limit to specific source(s)
`--wait`	—	off	Wait for completion inline
`--json`	—	off	Machine-readable output

# Hard quiz covering all sources
notebooklm generate quiz --difficulty hard --wait

# Easy quiz from specific sources
notebooklm generate quiz "Focus on chapter 3 concepts" --difficulty easy -s src_abc

# More questions, JSON output
notebooklm generate quiz --quantity more --json

Download the quiz after generation with notebooklm download quiz --format markdown.

flashcards — generate flashcards

Generate a flashcard deck for studying. Options match those for quiz.

notebooklm generate flashcards [description] [OPTIONS]

Option	Values	Default	Description
`--difficulty`	`easy`, `medium`, `hard`	`medium`	Card difficulty
`--quantity`	`fewer`, `standard`, `more`	`standard`	Number of cards
`-s, --source`	Source ID	all sources	Limit to specific source(s)
`--wait`	—	off	Wait for completion inline
`--json`	—	off	Machine-readable output

# Standard flashcards for all sources
notebooklm generate flashcards

# Harder cards from a specific source
notebooklm generate flashcards --difficulty hard -s src_abc --wait

# Fewer cards, JSON output
notebooklm generate flashcards --quantity fewer --json

infographic — generate an infographic

Generate a visual infographic summarizing your notebook sources.

notebooklm generate infographic [description] [OPTIONS]

Option	Values	Default	Description
`--orientation`	`landscape`, `portrait`, `square`	—	Image orientation
`--detail`	`concise`, `standard`, `detailed`	`standard`	Information density
`--style`	`auto`, `sketch-note`, `professional`, `bento-grid`, `editorial`, `instructional`, `bricks`, `clay`, `anime`, `kawaii`, `scientific`	`auto`	Visual style
`--wait`	—	off	Wait for completion inline
`--json`	—	off	Machine-readable output

# Landscape infographic with standard detail
notebooklm generate infographic --orientation landscape

# Detailed portrait infographic
notebooklm generate infographic "Key statistics" --orientation portrait --detail detailed --wait

# Kawaii style, JSON output
notebooklm generate infographic --style kawaii --json

data-table — generate a data table

Generate a structured data table from your notebook sources. The description is required and should specify what data to extract.

notebooklm generate data-table <description> [OPTIONS]

Option	Values	Description
`--wait`	—	Wait for completion inline
`--json`	—	Machine-readable output

# Generate a comparison table
notebooklm generate data-table "compare key concepts and their properties"

# Wait for completion
notebooklm generate data-table "list all studies with sample sizes and outcomes" --wait

# JSON output for scripting
notebooklm generate data-table "extract pricing data" --json

Download the result with notebooklm download data-table ./data.csv. The CSV uses UTF-8 with BOM encoding for Excel compatibility.

mind-map — generate a mind map

Generate a hierarchical mind map. This is the only synchronous generate command — it completes immediately and requires no --wait option.

notebooklm generate mind-map [OPTIONS]

Option	Values	Description
`--json`	—	Machine-readable output

# Generate a mind map (completes immediately)
notebooklm generate mind-map

# JSON output for automation
notebooklm generate mind-map --json

Download the mind map as a JSON tree with notebooklm download mind-map ./map.json. The JSON structure is {"name": "Topic", "children": [{"name": "Subtopic", "children": [...]}]}.

report — generate a text report

Generate a structured text report. Choose from built-in templates or write a fully custom prompt.

notebooklm generate report [description] [OPTIONS]

Option	Values	Default	Description
`--format`	`briefing-doc`, `study-guide`, `blog-post`, `custom`	`briefing-doc`	Report template
`--append TEXT`	Text string	—	Append extra instructions to a built-in template
`-s, --source`	Source ID	all sources	Limit to specific source(s)
`--wait`	—	off	Wait for completion inline
`--json`	—	off	Machine-readable output

# Study guide with default settings
notebooklm generate report --format study-guide

# Executive briefing from specific sources
notebooklm generate report "Stakeholder summary" --format briefing-doc -s src_001 -s src_002

# Append custom instructions to a built-in template
notebooklm generate report --format study-guide --append "Target audience: beginners"
notebooklm generate report --format briefing-doc --append "Focus on AI trends, keep under 2 pages"

# Fully custom report (description auto-selects custom format)
notebooklm generate report "Create a white paper analyzing the key market trends"

# Wait for completion, JSON output
notebooklm generate report --format blog-post --wait --json

--append adds instructions to the end of a built-in template prompt. It has no effect when --format custom is used; for custom reports, put your full prompt in the description argument.

Async behavior reference

Command	Async?	Notes
`audio`	Yes	Returns `task_id`. Audio can take 2–15 minutes.
`video`	Yes	Returns `task_id`. Video can take 5–30+ minutes.
`cinematic-video`	Yes	Alias for `video --format cinematic`.
`slide-deck`	Yes	Returns `task_id`.
`revise-slide`	Yes	Returns `task_id`.
`quiz`	Yes	Returns `task_id`.
`flashcards`	Yes	Returns `task_id`.
`infographic`	Yes	Returns `task_id`.
`data-table`	Yes	Returns `task_id`.
`mind-map`	No	Synchronous. Completes immediately.
`report`	Yes	Returns `task_id`.

After starting an async generation, poll with notebooklm artifact poll <task_id> or block with notebooklm artifact wait <id>.

Python Client

Types & Exceptions

CLI Reference

notebooklm generate — create AI content

Async behavior reference

Build docs developers (and LLMs) love

Python Client

Types & Exceptions

CLI Reference

Documentation Index

​Async behavior reference

Build docs developers (and LLMs) love

Async behavior reference