Google's NotebookLM can now search the web for you

Until now, using Google's NotebookLM came with one tedious chore: homework.

Before you could generate a flashy podcast overview or query your notes, you had to manually hunt down PDFs, copy URLs, and upload documents one by one. If you started with a blank slate, you had to do all the legwork yourself.

Google is changing that dynamic. The company announced Discover sources, a new feature for NotebookLM that searches the live web to find research materials for you.

Gemini becomes your personal web researcher

The workflow is straightforward. Inside any notebook, users can click a new Discover button in the Sources panel and describe any topic they want to explore.

Behind the scenes, Gemini combs through hundreds of potential web sources in seconds. It filters through the noise and presents up to 10 curated source recommendations.

Instead of just spitting out a list of links, each recommendation includes an annotated summary explaining exactly why that piece is relevant to your topic.

With a single click, you can import those recommended sources straight into your notebook. Once added, they work just like your manually uploaded files. You can turn them into Briefing Docs, generate Audio Overviews, ask follow-up questions in chat, and trace exact citations.

For anyone looking to test the feature without a specific prompt, Google also added an "I'm Feeling Curious" button that pulls sources on a random topic. The feature is rolling out to all users over the coming week, and Google teased that this is just the first in a series of Gemini-powered discovery tools heading to the app.

Why it matters

For students, writers, and knowledge workers, this eliminates NotebookLM's biggest bottleneck: the cold-start problem. Instead of spending an hour digging up starter reading on an unfamiliar topic, you get a vetted dossier in seconds, ready to be digested into study guides or audio briefs.

For the AI search market, NotebookLM is quietly turning into a serious rival for Perplexity and ChatGPT search. While those tools are great for answering questions on the fly, NotebookLM actually captures that research inside a persistent workspace where you can manipulate and synthesize it.

We think Google played this smartly by capping recommendations at 10 curated links with clear relevance annotations, avoiding the overwhelm of traditional search. The lingering risk, however, comes down to source quality. If Gemini pulls in biased articles or AI-generated spam from the open web, grounding your notebook on automated research could end up building projects on shaky foundations.