NextFin News - Josh Woodward, vice president of Google Labs, AI Studio and the Gemini app, joined Peter Yang on the Behind the Craft podcast for a live demo and wide‑ranging conversation about how Google ships AI products fast and what comes next for Gemini and NotebookLM. The episode was published on October 12, 2025 and features demos of Nano Banana, NotebookLM’s video overviews, Flow and internal developer tools used by Woodward’s teams. (listennotes.com)
The interview is presented as a demo‑forward discussion: Woodward shows examples of viral image outputs, talks through NotebookLM’s transformation features and explains the organizational practices that let small teams move from idea to people’s hands quickly. Peter Yang hosts the episode and guides the demos and conversation throughout. (listennotes.com)
Nano Banana: viral creativity and real‑world outcomes
Woodward opened the conversation by walking through Nano Banana use cases, emphasizing how users turned simple prompts into surprising creative and commercial outcomes. He described the phenomenon as having viral, regional origins — this actually started in Thailand
— and then spreading to other countries. He pointed to common patterns: style transfers, interior design placements and merchandise creations.
On the app’s home screen Woodward showed a default prompt for a miniature figurine and explained how users were able to upload an image
and generate a 1:17 scale figurine on a desk. He highlighted creators doing style transfers like watercolor effects, sellers placing their own artwork into generated interiors to sell prints, and physical merch such as stickers and embroidered designs derived from generated art.
"You put this kind of tool in people's hands and they go wild with it," Woodward said, describing how user creativity drove unexpected use cases.
He also described product evolution for Nano Banana: improved text handling in future models, better control for creators and tighter integration with animation and video so users can turn still outputs into motion. As he put it, expect more there
and the team is already working on next generations that give users more control and better handling of text on images.
NotebookLM: content transformations and video overviews
Woodward described NotebookLM as a product built to help people "understand anything" and highlighted a set of featured notebooks bundled with multiple sources. He demonstrated a NotebookLM workflow that turns a large set of sources into condensed, shareable outputs:
- Curated notebooks with many sources on the left and a Q&A and studio outputs on the right.
- Audio overviews and podcast‑style summaries that put NotebookLM on the map.
- Video overviews that create a 20–30 slide talking‑slide explainer from 70 sources.
Demonstrating the video overview, Woodward explained it would "pull out like a seven minute clip" and produce a slide deck that highlights key insights across a long report, enabling teachers, students and enterprises to turn a team's knowledge base into a compact explainer. He emphasized the power of these content transformations and the potential for consistent styling across outputs when combined with imagery models like Nano Banana.
Flow and rapid video generation
Woodward walked through Flow, a short‑form video creation tool showcased at Google I/O, and explained its rapid timeline: the project went from idea to people’s hands at Google I/O in under 100 days (he recalled it as about 86 days). Flow produces short, remixable clips (eight seconds by default) with audio, sound effects and the ability to insert reference images, then stitch clips into longer sequences in the Flow TV gallery.
He highlighted Flow’s remixability and control: users can open gallery examples, view the full prompt, bring in reference imagery, and direct animations and scenes. Woodward noted recent shipping cadence: the Flow team ships weekly or biweekly, added portrait mode for mobile use and optimized serving efficiency so power users can make many quick generations as part of a storyboard workflow.
How Google Labs builds a shipping culture
On organizational practice, Woodward described the ingredients that let small teams move fast inside a large company: tiny teams (often five to seven people), a premium on speed from idea to people’s hands and a willingness to staff up reactively after an initial product hit. He told the host, We put a huge premium on how fast you can go from idea to in people's hands.
He explained the team strategy in practical terms: start minimal to discover product‑market fit, avoid hiring large teams too early, and prioritize getting prototypes and early experiments into users’ hands so the product can be evaluated with real feedback. He reiterated that shipping early reveals how far refinement is needed and helps teams choose the right staffing and roadmap decisions.
Hiring signals and the Labs mindset
When asked what types of people thrive in Labs, Woodward pointed to people who tinker and build in their spare time: they express themselves through prototypes, not docs.
He described a short charter document the Labs team uses to explain who thrives there and listed key traits he looks for: rapid rate of learning, hands‑on tinkering, intellectual humility, kindness, high energy and optimism.
Reinforcing the culture point, Woodward said he values candidates who show side projects or GitHub work, deep product critiques, and people who are already active users and contributors to the community. His practical advice for applicants was to demonstrate what they build and their product thinking rather than depending solely on a resume.
Measuring early success: metrics and qualitative signals
Woodward described the metrics he uses in the early stages: small numerical milestones that demonstrate traction (for example, reaching the first 10,000 weekly or daily active users), combined with qualitative user observation. He warned that dashboards can mislead in early product stages and advocated going out to see users: go to a coffee shop, go to a university campus
and watch people use the product, because retention and delight are the best proxies for usefulness in early experiments.
Gemini’s future: personal, proactive, powerful
Throughout the interview Woodward framed Gemini’s next phase with three Ps: personal, proactive and powerful. He said personal context is a primary focus and that the team is experimenting internally with connecting Gmail, Google Photos and other Google data sources to give Gemini deeper personal context — with an emphasis on user control, clear permissions, privacy and security.
He described how a personal Gemini could surface useful, timely suggestions: wake up in the morning and it's like hey Peter here's your seven meetings today. Here's two or three things in each meeting that are really important.
Woodward stressed that the multimodal nature of Gemini — text, images, video and audio — is a differentiator and enables editing and conversational editing scenarios like Nano Banana and VO that other models find difficult.
Internal tooling and co‑creation with AI
Woodward reviewed internal tools that accelerate product work: AI Studio for rapid prototyping and iteration, Opal as a visual node‑based editor for chaining calls, and coding assistants like Jules that autonomously generate fixes and PRs. He gave examples of people rebuilding Flow prototypes in AI Studio in a week and lawyers using Opal workflows to pre‑review product documents for legal issues.
On coding and agentic workflows he described practical loops: run a bug bash, feed issues to Jules and get PRs or suggested fixes back — a pattern that illustrates co‑development with AI and faster iteration across creative and engineering workflows.
Interaction models: beyond typing
Discussing future interaction modes, Woodward said he expects the interface landscape to diversify beyond typed chat. He noted progress in voice understanding and generation and multimodal inputs, and observed that live spoken conversations with Gemini can be considerably longer; people may prefer voice while walking or driving and use typed chat in noisier or public settings. In his words: I think we're in the early innings of this — there's tons of UX and UI innovation coming.
Closing and where to follow Josh
Woodward closed the conversation by encouraging feedback and community engagement. He said he is active on X (formerly Twitter) at @joshwoodward and participates in the Labs Discord for product feedback. The episode host is Peter Yang and the conversation is published on the Behind the Craft podcast. (listennotes.com)
References
Episode page and transcript: Behind the Craft — Inside Gemini and NotebookLM. (listennotes.com)
Spotify episode listing: Inside Gemini and NotebookLM | Behind the Craft. (creators.spotify.com)
Additional episode listing: GetPodcast — Inside Gemini and NotebookLM. (getpodcast.com)
Find Josh Woodward on X: @joshwoodward.
Explore more exclusive insights at nextfin.ai.
