Behind the product: NotebookLM | Raiza Martin (Senior Product Manager, AI @ Google Labs)

The conversation opens with a live demonstration of NotebookLM’s Audio Overview: a user supplies a source and the product generates an AI audio discussion. Lenny sees its rapid spread as evidence that people are rethinking what AI can do and how they might engage with information. Raiza describes the audio format as an accessible way to experience material that would otherwise remain text.

For Raiza, the experience depends on more than converting text to speech. She points to the hosts’ laughter, interruptions, verbal reactions, and changing intonation as features that make the generated conversation feel engaging. The team’s early experiments were less effective, so it listened repeatedly and worked to shape the model’s behavior and the product experience around it.

The idea grew from a product question: users could already interact with material through text, but some wanted an audio response. Raiza says she personally liked voice interaction and found that audio output changed how she engaged with technology and thought in the moment. The team therefore looked for a format that made the feature’s value easy to experience and enjoyable to try.

Notebook began as a Google Labs project exploring how a language model could help people interact with a focused set of source material. Raiza was leading AI Test Kitchen; the early work involved her, an engineer, and later Steven Johnson. She recalls that the engineer was the only person fully dedicated to the technology, while other contributors joined the project alongside their regular work.