Fresh off impressive releases at Google’s I/O event, three Google Labs leaders explain how they’re reimagining creative tools and productivity workflows. Thomas Iljic details how video generation is merging filmmaking with gaming through generative AI cameras and world-building interfaces in Whisk and Veo. Jaclyn Konzelmann demonstrates how Project Mariner evolved from a disruptive browser takeover to an intelligent background assistant that remembers context across multiple tasks. Simon Tokumine reveals NotebookLM’s expansion beyond viral audio overviews into a comprehensive platform for transforming information into personalized formats. The conversation explores the shift from prompting to showing and telling, the economics of AI-powered e-commerce, and why being “too early” has become Google Labs’ biggest challenge and advantage.

Hosted by Sonya Huang, Sequoia Capital

00:00 Introduction

02:12 Google's AI models and public perception

04:18 Google's history in image and video generation

06:45 Where Whisk and Flow fit

10:30 How close are we to having the ideal tool for the craft?

13:05 Where do the movie and game worlds start to merge?

16:25 Introduction to Project Mariner

17:15 How Mariner works

22:34 Mariner user behaviors

27:07 Temporary tattoos and URL memory

27:53 Project Mariner's future

29:26 Agent capabilities and use cases

31:09 E-commerce and agent interaction

35:03 Notebook LM evolution

48:26 Predictions and future of AI

Mentioned in this episode: 

Whisk: Image and video generation app for consumers

Flow: AI-powered filmmaking with new Veo 3 model

Project Mariner: research prototype exploring the future of human-agent interaction, starting with browsers

NotebookLM: tool for understanding and engaging with complex information including Audio Overviews and now a mobile app

Shop with AI Mode: Shopping app with a virtual try-on tool based on your own photos

Stitch: New prompt-based interface to design UI for mobile and web applications.

ControlNet paper: Outlined an architecture for adding conditional language to direct the outputs of image generation with diffusion models