Three developers who helped build Google's viral AI research tool NotebookLM just launched their own startup with a $4.6 million funding round. Huxe promises to transform how people consume information by turning any topic into an interactive AI podcast, complete with multiple hosts you can interrupt and question in real-time.
The NotebookLM exodus is paying off. Three core developers who worked on Google's breakout AI research assistant since day one just secured $4.6 million to build something even more ambitious - an audio-first platform that turns information consumption into interactive conversations.
Raiza Martin, Jason Spielman, and Stephen Hughes left Google last December with a clear vision: people were obsessed with NotebookLM's podcast feature, but they wanted something more personal and immediate. Their answer is Huxe, which launched publicly this week after months of invite-only testing.
The funding round attracted some serious names. Google Research's chief scientist Jeff Dean backed his former colleagues, along with Conviction, Genius Ventures, and Figma CEO Dylan Field. It's rare to see Google's top AI researcher invest in a startup built by his own former team - a signal of just how promising this space looks.
"During this phase, we realized that people liked having the ability to generate audio for different topics," Martin told TechCrunch. "We also observed that people often used the app at specific times to get their daily brief or catch up with news while getting ready."
That behavioral insight drove everything. Unlike NotebookLM's document-focused approach, Huxe connects directly to your email and calendar to generate personalized daily briefings. But the real innovation comes with what they call "live stations" - persistent audio feeds that track developing stories across multiple sources and update you as news breaks.
Think of it as AI-powered radio, but smarter. You can create a station for "tech acquisitions" or "climate policy" and Huxe will continuously monitor relevant sources, generating fresh podcast segments as the story evolves. The AI hosts don't just read updates - they synthesize changes, highlight contradictions between sources, and explain why developments matter.
The interactive element sets Huxe apart from both traditional podcasts and static AI summaries. Users can jump into any conversation to ask follow-up questions, request different perspectives, or dive deeper into specific points. The AI hosts respond naturally, maintaining the conversational flow while adapting to your interests.
