Google hit back at its AI competitors with an impressive series of announcements and new products at its annual developer conference in Mountain View, California. The most direct impact for media and marketing will be a powerful new video generator, further integration of AI into the basic search experience, and a standard language model packed with agent integration.
Although it’s only been a few hours since it was announced and the conference is still underway, Mumbrella took a look at the release and spoke to our contacts to get a sense of what’s to come.
Shining: Gemini Omni
So many bubbles: Omni-generated video still example (Google)
Google’s wording on the new generator Gemini Omni is bold: it will eventually allow you to “create anything from any input.”
advertisement
For now, you’ll have to settle for generating video from text, image, and video input. Examples provided in the presentation show that Omni can change camera angles, remove objects, change the background, and edit videos in natural language.
Vin Schifferstein-Vidal, co-founder of AI video production company MC&V, already has a good understanding of Omni’s capabilities, saying it is comparable to ByteDance’s SeaDance video generator.
“It’s on par with Seedance in terms of output, but perhaps more inclusive of different styles. Seedance is very good at photorealism, but not as strong in areas like character development and stylized formats like animation.”
Google says Omni is “based on Gemini’s real-world knowledge.” This seems to fall short of a perfect world model, but may prevent obvious physical absurdities from being accidentally generated.
“Physics is [of Omni] It’s definitely improved,” Schifferstein said.
Omni is immediately available to Google’s top end of AI users: video production tools Flow and AI Plus, Pro, and Ultra subscribers within the Gemini app.
“From a workflow perspective, its integration into the flow makes it much easier to maintain consistency between shots. The agent feature is probably the biggest novelty here, making the experience much more intuitive for users. You can move back and forth within a shot, adjust moments, and iterate without completely changing the shot itself.”
“More importantly, this is driving more multi-shot agent filmmaking, and it’s clearly where we’re all headed. That said, we’re still in the early stages in terms of testing.”
Important point: Changes to the search interface
New search interface (click to expand)
Google continues to further integrate its search engine and AI products, announcing new AI-powered updates to Google Search.
Until now, AI has only appeared in Google Search as AI Overview and as a separate AI mode similar to conversations with chatbots. Google’s revamped search engine is powered by the new Gemini 3.5 Flash model (see below) and represents a major shift towards AI.
Paul Hewett, CEO of In Marketing We Trust, said the changes were “the biggest redesign of the search box in 25 years, as well as a significant change to the information agent”.
“This will impact how search works,” he says.
“AI Overview and AI Mode are now cascaded with context preserved. Instead of being cited once, units of brand visibility are cited and surfaced in subsequent follow-ups.”
“This is nice.”
The revamped search field feature allows for longer, more complex inquiries that feel more conversational in the form of new “intelligent search boxes” like the Gemini chatbot. The move follows Google’s move from AI Overview to AI Mode and now comprehensive AI Search. The new engine also accepts uploads in the form of PDFs and photos, automatically generates subtle prompts, and agentically accesses and monitors research topics on its own.
Google automatically creates visual and lightweight apps (called “widgets”) in response to specific prompts. For example, create a trip planner that combines your current location, upcoming calendar events, and real-time criteria to create a personalized plan.
Transitions between AI Overview and AI Mode are also streamlined, allowing you to talk to the AI that provided your search results to get the answers you need. Google also describes a move into an “agent” era where AI assistants can help book services, send alerts, and monitor live events. These features won’t be available until later this year.
Solid: Gemini 3.5
Starting today, Google’s base-level AI experience will be powered by a new model, Gemini 3.5 Flash, which is clearly just one of a series of Gemini 3.5 models to come. Google’s AI branding turmoil continues.
Google says Flash is built with agent tasks in mind. According to the data provided in the presentation, Flash beats some benchmarks, but beats Anthropic’s Claude and OpenAI’s ChatGPT 5.5 in some others.
Google’s own benchmark data (click to expand)
Google claims that the big advantage of Flash is speed (and therefore probably cost). “3.5 Flash delivers frontier-level intelligence at exceptional speed,” which makes it excellent for “long-term agent tasks,” he said.
Shopping powered by AI
Announced at the conference, Google’s Universal Cart was introduced as a unified shopping experience that combines multiple merchants into a single, streamlined shopping experience.
The foundation is Google’s Shopping Graph, which the company bills as the world’s most comprehensive product directory, with over 60 billion up-to-date product listings.
The system promises to track price changes over time, notify you when a product is back in stock, and flag compatibility issues between purchases (if you accidentally add an incompatible accessory to your device at the same time).
Suresh Ganapathy, senior director of consumer shopping at Google, told reporters before the conference that he wants to make shopping more fun.
“We keep hearing from shoppers that they really enjoy the fun parts of shopping, but would rather leave the tedious parts to AI,” he said.
Developed in collaboration with online retailers, the shared language UCP enables agents and shopping sites to work together across the entire buyer journey. Another tool, Agentic Payment Protocols (or AP2), allows agents to make purchases on behalf of shoppers.
Paul Hewett believes Universal Cart has the potential to change the face of retail.
“The press doesn’t reflect that yet, but it’s probably strategically so…Karting is the visible part.
“Under Universal Cart and UCP and AP2, Google provides the infrastructure to own the entire shopping journey: discovery, recommendations, checkout, payments, and fulfillment. It’s all within the surface of Google, and Google has the data for every step. It’s scary…”
Hewett said Google may have “deliberately underestimated it.”
“Google already knows what you search for, click on, watch, and where you go. Add Universal Cart and it knows what you were thinking of buying. Add AP2 and it knows what you actually bought, from whom, and for how much. The top inference layer can predict intent and price sensitivity at a level that retailers can’t match with their own data. We need to have smart conversations about this.”
Everything else
No more glass holes: the new glasses were designed by The Gentle Monster (left) and Warby Parker (right).
There are also many new announcements. Google is expanding its capabilities AI content detector Beyond the Gemini app, you can now find AI-generated watermarked videos, images, and audio in Google Chrome and Google Search. The feature was first released late last year exclusively on the Gemini app.
Youtube’s search functionality has also been upgraded. A new feature called Ask a question on Youtube We’re bringing Google’s AI-powered search capabilities to our video platform, promising users the ability to ask more complex queries and get more specific results. Ask Youtube also allows you to cut out the exact section of the video that answers a specific query. This feature is currently available for premium users in the US and will be widely rolled out soon.
document live is a new feature in Google Docs that allows users to convert audio recordings into a cohesive stream of generated text. If you have access, you can also narrow your results by looking at connected Google Accounts and the web.
intelligent eyewear This is Google’s answer to the Meta Ray Ban (and more), created in partnership with Samsung. Google provided the software and Samsung provided the hardware. These are available for both iOS and Android, but the price remains a mystery for now.
project genie is an experimental AI system that allows users to generate and explore interactive 3D environments from simple prompts, and has now been updated to incorporate real-world imagery from Street View.
The integration, which allows users to create virtual scenes based on real-world locations and creatively modify or reimagine them to create short, explorable simulations that blend AI-generated content with real-world geographic data, is being rolled out to premium subscribers in stages, with broader expansion planned.
