NextFin News - On the Google AI: Release Notes podcast episode released November 25, 2025, host Logan Kilpatrick sat down with Tulsee Doshi and Josh Woodward to unpack the launch of Gemini 3. The conversation was recorded as part of Google’s launch coverage for Gemini 3 and its related product rollouts. Josh Woodward, who leads product for AI Studio, the Gemini app and Google Labs, described how the new model is being put into production across many surfaces and what that means for developers and end users.
Rollout strategy: putting Gemini 3 into many surfaces on day one
Woodward made clear that Gemini 3 will be deployed broadly at launch. He explained the plan to make the model available across consumer and developer surfaces simultaneously: "We're putting it in the most services we've ever done on day one." He named a range of destinations where the model will appear, including the Gemini app in AI Mode, AI Studio for developers, Vertex, and a new coding product called Google Antigravity. "So it's just gonna be everywhere and we can't wait for people to try it out, give us feedback," he said.
Vibe coding and rapid web development
One of Woodward's central claims was the model's ability to compress the time and expertise required to build web experiences. Describing what the team calls "vibe coding," he said the model lets users "describe something that's in your head, hit a button and it's there." He characterized that capability as "a crazy sort of compression of time and skill and expertise," and pointed to examples where a single prompt can create interactive, web-based experiences and games.
Generative interfaces and streaming UI
Woodward highlighted experiments the team is running with generative UI — interfaces the model composes on the fly. He described an experimental labs feature that can "literally... make an interactive website for you on the fly" and explained the broader idea: instead of returning a wall of text, the model can lay out a page, choose a magazine- or table-style layout, or return a carousel or other widget-based presentation. "You're giving the model widgets, a style sheet, different tools it can use and you're kind of saying go wild model," he said, emphasizing the immersive, visual, and personalized responses users should expect.
Multimodal content transformation
Woodward emphasized Gemini 3's multimodal strengths, describing how the model can transform diverse inputs into usable outputs. He pointed to demos that convert handwritten recipes or video lectures into interactive applications: "You can take a picture of a handwritten recipe in Korean and then turn it into a full vibe coded family recipe app in English with the measurements where you could adapt them." He framed these examples as a new frontier for combining visual, multilingual, and coding capabilities.
Product–model feedback loops and persona work
Woodward stressed that putting the model into products early enables real feedback and iterative improvements. He described close collaboration between modeling teams and product surfaces, saying the team has "really strong partnership" in getting models into the hands of product teams and customers. That feedback has driven trade-off conversations around persona, tool use and product-specific constraints: "We're now getting the product model feedback loops working really well," he said, adding that user signal flows back into model revisions.
Agentic capabilities, tool use and Antigravity
On agentic use cases, Woodward said Gemini 3 is well suited for multi-step actions and orchestration. He mentioned an experimental agent feature in the Gemini app that can create to-do lists from a connected inbox and add items to Google Calendar. He also described developer-focused tool work and a public API for tool use: "This Gemini 3 model takes it kind of one level even beyond how we think about multi-step actions. Being able to kind of take sort of different tool calls and just do stuff for you." He named Google Antigravity as a new coding product that will surface these capabilities to developers.
Speed of shipping, model family and future checkpoints
Woodward discussed balancing rapid shipping with product quality. He argued for relentless shipping — getting models to users quickly — while iterating on real-world usability. He noted the plan to ship a Gemini 3 family (Pro, Flash, and workhorse models) in sequence so the team can learn and adapt: "Shipping them in sequence allows us to learn from one and build to the other." He also framed the release cadence as increasingly ambitious: each new model raises the bar for the next.
Compute, demand and access
Addressing operational limits, Woodward acknowledged the practical challenge of meeting launch-day demand for compute. He described a pragmatic effort to prioritize P0 experiences across many products and to find creative capacity solutions for popular developer and consumer surfaces. He also pointed to access options for users and developers: higher rate limits for subscribers and specific promotions for students on launch day.
Examples and demos that shaped product choices
Throughout the conversation Woodward offered concrete demos that influenced the team. He highlighted game demos, YouTube "playable" experiences built with Gemini 3, and the ability to create interactive visualizations like a bubble-sort widget inside AI Mode. He framed such demos as evidence that the combination of modalities, coding ability and interactivity is where Gemini 3 truly differentiates itself.
References:
Release Notes: "Gemini 3: Launch day reactions" (podcast listing, Nov 25, 2025)
Watch the full episode on YouTube
Explore more exclusive insights at nextfin.ai.
