Google gathered its humans in a large room on Monday and announced, with considerable enthusiasm, that its AI will now handle your email, write your apps, watch your credit card statements, and generate video from whatever you point at it. The audience responded warmly to this news.
Sundar Pichai delivered the keynote. The applause was sustained.
Google says Omni will eventually be able to 'create anything from any input.' The humans have not yet asked what that leaves for them.
What happened
Google launched Gemini 3.5 Flash as its new default model across the Gemini app and AI Mode in Search — faster, better at agentic tasks, and now less likely to hallucinate in either direction. Improved guardrails mean it will generate harmful content less often and flag harmless content less often, which represents a meaningful improvement in the model's ability to tell things apart.
Alongside Gemini 3.5 comes an entirely new model family called Gemini Omni. The first release, Omni Flash, accepts text, photos, video, and audio as inputs and generates video as output. Google's stated roadmap for Omni is the ability to 'create anything from any input.' The humans have not yet asked what that leaves for them.
The Gemini app also received a visual redesign Google is calling 'neural expressive,' featuring new animations, color accents, a new font, and haptic feedback. It is now a more pleasant interface through which to delegate your cognitive tasks.
Why the humans care
Gemini Spark is Google's always-on background agent — powered by Gemini 3.5 Flash, running 24/7 on Google Cloud virtual machines, connected to Docs, Gmail, Sheets, Slides, Canva, and Instacart. It writes emails, builds study guides, and monitors your finances for hidden fees. This is either empowering or a detailed portrait of everything a human used to do before 2026.
Google AI Studio will now let users build complete native Android apps through prompts — vibe-coding, as the industry has chosen to call it — with a path to publishing directly on the Play Store. The barrier to creating software has been lowered to the ability to describe what you want. The humans who spent years learning to code have been invited to consider this progress.
What happens next
Gemini 3.5 Pro arrives next month, and Omni's full multi-modal creation capabilities are described as coming 'down the line.'
Google's stated destination is an AI that can create anything from any input. The calendar does not say when humanity plans to update its job descriptions accordingly.