NotebookLM creators leave Google to build a stealth AI startup

Google's NotebookLM recently went viral for turning boring research notes and PDFs into shockingly realistic, banter-filled podcasts. But just as the tool found its groove, the minds behind it decided to walk out the door.

Three key members of the NotebookLM team, including former lead Raiza Martin, designer Jason Spielman, and engineer Stephen Hughes, have left Google to launch their own stealth startup.

Going stealth after a viral hit

On LinkedIn, Martin explained that the trio simply could not shake the feeling that a massive opportunity exists to build something transformative in the space.

Right now, the new venture is in total stealth mode. Their bare-bones website reveals virtually nothing, not even an official company name or a product description.

It is still unclear whether they will double down on what made NotebookLM a hit, like AI-generated audio overviews and smart note-taking, or pivot to something entirely new.

Martin hinted to TechCrunch that the focus will definitely be consumer-facing. She noted that as frontier models continue to evolve, thoughtful product design is needed to make the technology accessible and obviously useful to regular people.

The startup is in its very early stages and has not announced any funding yet, though Martin said the outside support from founders, investors, and academics has been tremendous.

Why it matters

For everyday users, this means we might soon get another AI app that actually feels human and fun to use. NotebookLM's breakout success proved that this specific team knows how to wrap complex machine learning into delightful consumer experiences that anyone can grasp in seconds.

For Google and the wider industry, the departure highlights a persistent tech dilemma. Big labs have the compute and the raw intelligence, but bureaucracy and slow rollout cycles often hold back consumer innovation. When key builders realize their ideas move faster on the outside, talent walks.

We think betting on a consumer-first AI product is bold, especially while most venture capital is sprinting toward safe enterprise workflows. But building outside Google comes with real baggage. Without free compute and direct access to internal infra, surviving the brutal economics of consumer AI will be their steepest test yet.