Subvolts take Google DeepMind reveals how user feedback is driving Gemini 3 and Nano Banana Pro development. Jan 17, 2026 - Developers
- Learn about user-driven feature development and how real-world usage influences AI model roadmaps at Google DeepMind.
- Business
- Understand how unexpected AI applications and user demand are shaping product strategy and future AI capabilities.
- Everyday users
- Discover how your usage of AI tools like Gemini and image editors might influence upcoming features and improvements.
This interview provides direct insight from a Google DeepMind product leader on current AI developments and user feedback.
Google DeepMind's Tulsee Doshi: Gemini 3, Deep Think's Unexpected Uses, Layer-Based Editing
Audio streams from Jason Howell and Jeff Jarvis. Your place is remembered on this device.
From Subvolts
Summary and analysis by Subvolts editorial, AI-assisted.
Google DeepMind's Tulsee Doshi discusses how real-world AI usage shapes Google's roadmap on AI Inside. The episode covers Gemini 3's reasoning advances, unexpected creative uses of Deep Think, and user demand for layer-based image editing in Nano Banana Pro. Doshi also shares insights on internal adoption and Google's success at the International Mathematical Olympiad.
This Google AI podcast talks about how people are actually using their AI tools, like Gemini 3. They learned that one tool is surprisingly good at writing stories, not just research, and users want more control over editing images, like in Photoshop.
This episode offers a look into how Google DeepMind is adapting its AI strategy based on actual user interactions. It's for anyone interested in the practical application of AI models, particularly those curious about Gemini 3's capabilities and the future direction of image editing tools. The discussion on user demand and unexpected use cases provides insight into the current consumer-facing AI landscape and how labs are responding to real-world feedback.
- HostJason Howell
- HostJeff Jarvis
- GuestTulsee DoshiSenior Director and Head of Product for Gemini Models, Google DeepMind
- Guest: Tulsee Doshi, Senior Director and Head of Product for Gemini Models at Google DeepMind
- Show: AI Inside
- Length: 37 minutes
- Gemini 3: Features reasoning breakthroughs.
- Deep Think: Excels at creative writing.
- Nano Banana Pro: Pursuing layer-based editing due to user demand.
This episode is sponsored by Your360 AI. Get 10% off through January 2026 at Your360.ai with code: INSIDE.
Tulsee Doshi is Senior Director and Head of Product for Gemini Models at Google DeepMind. She joins us to reveal how real-world usage is reshaping Google's AI roadmap in unexpected ways. We explore the surprising discovery that Deep Think excels at creative writing rather than just academic research, why the team is now pursuing layer-based editing capabilities for Nano Banana Pro based on user demand, and how internal adoption patterns exceeded expectations. We also cover Gemini 3's reasoning breakthroughs, the path from research to production, and what Google learned from achieving gold medal performance at the International Mathematical Olympiad. Hit play for a candid discussion about AI development.
Note: Time codes subject to change depending on dynamic ad insertion by the distributor.
CHAPTERS:
00:00:00 - Podcast begins
00:01:32 - Introducing Tulsee Doshi, Head of Product for Gemini Models
00:03:20 - Nano Banana Pro Launch: Surprises and User Adoption Two Months Later
00:04:17 - What Users Are Asking For: Future Directions for Nano Banana Pro
00:05:21 - User Demand for Photoshop-Style Editing and Layered Image Control
00:06:04 - Gemini 3 Integration Across Google Products and Beyond
00:08:23 - Will AI Change How We Interact With Google Products?
00:12:20 - The Future of Real World Models and 3D Spatial Understanding
00:27:36 - Deep Think Explained: When to Use Different Gemini Modes
00:28:41 - Deep Think Use Cases: Academic Research, Enterprise, and Consumer Applications
00:30:51 - Surprising Deep Think Application: Creative Writing Beyond Academic Research
00:31:45 - The Future of AI-Generated Movies and Long-Form Content
00:32:00 - Thank you to Tulsee Doshi for joining the AI Inside podcast
Learn more about your ad choices. Visit megaphone.fm/adchoices
- Deep Think
- Tulsee Doshi
- A Google AI tool.
- Nano Banana Pro
- Tulsee Doshi
- A Google AI product, likely for image editing.
- Layer-based editing
- Tulsee Doshi
- A way to edit images by stacking different elements or adjustments on separate layers, offering more control.
1 h 23Is AI Really Bringing the End of Software?
WithBrandon Thomas
50 minQualcomm Wants Its AI Chips On Every Device
WithJohn Solomon
Same guests, other shows
32 minGoogle I/O 2026 Recap with Logan Kilpatrick, Josh Woodward and Tulsee Doshi
WithJosh Woodward, Tulsee Doshi
59 minThe Model Eats the Scaffolding: DeepMind's Logan Kilpatrick & Tulsee Doshi on 3.5 Flash, Omni & More
WithLogan Kilpatrick, Tulsee Doshi
40 minGoogle I/O 2025 Recap with Josh Woodward and Tulsee Doshi
WithJosh Woodward, Tulsee Doshi
28 minLaunching Gemini 2.5
WithTulsee Doshi
56 min
45 minHow Mintlify Is Rebuilding Documentation for Coding Agents
WithHan Wang, Hahnbee Lee
44 minInferact: Building the Infrastructure That Runs Modern AI
WithMatt Bornstein, Simon Mo, Woosuk Kwon
Questions people ask
- What is Deep Think and how is it being used?
- Deep Think is a Google AI tool that is surprisingly excelling at creative writing, contrary to initial expectations of it being solely for academic research.
- What new features are coming to Nano Banana Pro?
- Based on user demand, Google is pursuing layer-based editing capabilities for Nano Banana Pro, similar to Photoshop, to give users more control over image manipulation.
Page generated Oct 19, 2026. Summaries are Subvolts' own; the audio, artwork and notes belong to Jason Howell and Jeff Jarvis.