Google opened its annual developer conference with a clear message – the era of passive AI chatbots is ending. Under CEO Sundar Pichai, the company unveiled its most ambitious push yet to make Gemini a proactive, always-on intelligence layer woven into Search, productivity tools, hardware, and daily life.
The event centered on agentic AI, i.e., systems that don’t just respond but anticipate needs, handle multi-step tasks in the background, and create across text, images, video, and audio. Google also doubled down on wearable AI with stylish smart glasses powered by Android XR, taking on Meta smart glasses and Apple’s upcoming rumoured smart glasses, also signalling a future where computing becomes more ambient and less screen-bound.
Here’s an expanded breakdown of the major announcements:
Gemini AI model and Core updates
Gemini 3.5 Flash: Google’s fastest and most capable agentic model yet, delivering frontier-level performance in coding, tool use, and multi-turn reasoning at improved speed and lower cost. The model is available across the Gemini app, Google Cloud, and developer platforms, with a Pro variant expected soon.
Gemini Omni and Omni Flash: A new multimodal family designed to “create anything from any input.” It shines in video generation and editing, i.e., users can remix scenes, add characters or effects, change environments, or generate multiple camera angles from a single photo or clip. Broader any-to-any capabilities are planned. Rolling out with enhanced “Neural Expressive” features.
Gemini Spark: The highlight for many is a 24/7 proactive personal agent that runs in the background on dedicated cloud resources. It organises emails, prepares briefs, monitors topics, coordinates events, and executes complex tasks across Google apps even when devices are offline. It will be initially available to testers and Google AI Ultra subscribers, with wider rollout planned.
Gemini app redesign: Fresh UI with fluid animations, a prominent Ask Gemini bar, and improved Gemini Live (faster responses, better noise handling, expanded dialects).
Gemini for Science: New research tools for summarising papers, generating hypotheses and code, and accelerating breakthroughs in medicine and weather modelling.
Mac integration: Voice-powered features let users dictate naturally (with “ums” and corrections handled) and have Gemini organise files or turn thoughts into polished documents.
Pricing tweaks: Adjustments to higher-tier plans, including reductions for Ultra subscribers.
Hardware and XR (Smart Glasses Push)
Google’s biggest hardware story in years since the Pixel smartphones! These stylish audio-first smart glasses (partnering with Samsung, Warby Parker, and Gentle Monster) will launch this fall. They offer hands-free Gemini assistance, cameras for context, real-time translation, navigation, private audio, and app actions — all compatible with Android and iOS.
Advanced AR/display versions are in development for 2027. Demos showed seamless tasks like ordering coffee without pulling out a phone.
Search and Discovery
– Persistent Information Agents: Users can now spin up custom AI agents from the Search box that continuously monitor topics (stocks, product drops, news) and deliver proactive alerts.
– Ask YouTube: Quickly surfaces exact moments in long videos that answer queries.
– Enhanced AI content provenance tools (SynthID + C2PA) and generative UI features that build custom interactive tools on the fly.
Productivity and Creative Tools
Google Flow: Deeper agentic workflows and large-scale edits, now supercharged by Omni models for advanced video creation.
Universal Cart: An AI-powered shopping assistant across Google services that tracks deals, compatibility, and alerts.
Google Pics and Docs/Gmail Live: New image tools for Workspace plus natural dictation that turns rambling speech into structured, polished content.
Stitch enhancements for natural-language collaborative design.
Android and Ecosystem
– Reinforcement of Android 17 features (many previewed earlier), including Gemini Intelligence layer and on-device agentic tools.
– Revealing the Googlebook laptops running the new OS with the Magic Pointer, plus broader ecosystem integrations like Android Auto.
Other Highlights
Project Genie: AI world models grounded in real-world data (e.g., Street View) for realistic yet imaginative experiences.
CodeMender: A new code security agent that automatically finds and fixes critical software vulnerabilities. Google is inviting a select group of experts to test the new CodeMender API, with a broader launch coming soon. This underscores Google’s focus on addressing AI-related security risks while advancing the technology.