Google's source-grounded research tool that now turns your uploaded documents into TikTok-style vertical video summaries.
Last verified · 2026-07-01 · by Moe Ameen
NotebookLM is Google's AI research and note-taking tool. You upload your own sources — PDFs, Google Docs and Slides, web links, YouTube videos, pasted text, or audio — and it becomes a research assistant grounded in that material. Every answer, summary, and generated output is built only from what you gave it and is cited back to the source, which is the whole point: it is a summarizer of your documents, not a general chatbot that invents facts.
Its best-known outputs are Audio Overviews (a two-host, podcast-style discussion of your sources) and Video Overviews. The video side has grown into a family: landscape "Explainer" and cinematic formats aimed at longer, narrative explainers, and — added in mid-2026 — Short Video Overviews, a roughly 60-second vertical clip built for phones. Each Short Video Overview narrates a single concept from your notebook over animated visuals, and Google framed it openly as "doom scrolling but make it educational."
The vertical clips are generated using Nano Banana 2 Lite, Google's fast, low-cost image model (officially Gemini 3.1 Flash-Lite Image). The demo look leans on stylized, paper-cutout-style animation rather than photorealistic AI video, which deliberately sidesteps the uncanny feel of realistic generated footage. Access has rolled out across Google's paid AI Pro and AI Ultra tiers with wider availability following, and the video features started English-only, so treat exact availability, languages, and limits as a moving target.
NotebookLM is a research-to-media tool, not a publisher. It hands you one clip, in one of its own preset house styles, summarized from your sources — there is no brand voice, no persona, no aspect-ratio control beyond the format presets, and no scheduling or posting. Getting that output on-brand and onto platforms is a separate job.
NotebookLM is strongest as the front of your pipeline: the place you dump raw research and get back a cited, de-duplicated understanding of it — plus one summary clip to sanity-check the story. Where it stops is exactly where a posting schedule begins. It gives you a single video in NotebookLM's preset paper-cutout style, summarized from your sources, that you download and place yourself, on the tiers and timeline Google decides. That is a great study aid; it is not a content operation.
Kompozy takes the same source and makes the shape repeatable and on-brand. Feed the research (or the transcript of a NotebookLM Audio Overview, or your own take on what it surfaced) into Kompozy and it generates the vertical explainer as a Listicle Video or a Persona Short — with AI narration, auto-captions, and your aspect ratio and brand styling through HyperFrames, instead of NotebookLM's fixed house look. Then it fans the same source into the formats NotebookLM never touches: a carousel, a quote graphic, a blog draft, an email newsletter, and platform-native captions written in your voice via the Persona Brief — and schedules and publishes the whole set across all nine connected platforms. NotebookLM distills the knowledge; Kompozy turns one distilled source into a week of branded, scheduled, cross-platform content. That research-to-short-form-video workflow is the exact pipeline covered in the /guides/ai-research-to-short-form-video playbook.
They are Video Overviews — NotebookLM turning your uploaded sources into a narrated video. The newest format, Short Video Overviews, is a roughly 60-second vertical clip on a single concept with animated visuals, generated by Nano Banana 2 Lite. Longer landscape Explainer and cinematic formats also exist. Every video is grounded in and cited to the material you upload.
NotebookLM has a free tier for its core research and note-taking. The newer video features, including Short Video Overviews, rolled out first on the paid Google AI Pro and AI Ultra tiers with wider access following. Availability, limits, and supported languages are still settling, so confirm current access in NotebookLM.
No. NotebookLM generates a video from your research that you download or share yourself. There is no scheduling, no multi-platform publishing, no brand voice, and only its own preset visual styles. To turn a source into an on-brand vertical clip plus carousels, captions, and posts published across nine platforms, pair it with a content engine like Kompozy.
NotebookLM is a source-grounded research tool that summarizes your documents into audio and video. Kompozy is a content generation and publishing engine: it produces on-brand persona/avatar video, listicle and clipped shorts, carousels, images, blogs, and newsletters from a source, then schedules and publishes them across nine platforms. Use NotebookLM to research and distill; use Kompozy to produce and ship.
Only within NotebookLM's preset formats and styles (such as the demo's paper-cutout animation). There is no full brand-styling control, custom aspect ratios beyond the format presets, persona, or logo lockup. For pixel-exact brand styling and your own aspect ratios, you render the vertical format in a tool like Kompozy using HyperFrames instead.