Google I/O 2026 kicked off with a focus on artificial intelligence, as the company unveiled a slew of new products, features, and updates across its ecosystem. The keynote, held on May 19, was dominated by Gemini upgrades, but also included major news about Android XR, Wear OS, Google Search, and even a glimpse at a new AI lab project called Sophie. Here's a comprehensive look at the biggest takeaways from this year's developer conference.

Gemini Expands with New Models and an Always-On Agent

Gemini was once again front and center. Google introduced the Gemini Omni family of models, starting with Omni Flash, which can generate videos from text, images, audio, and video inputs. The company envisions Omni as a tool to "create anything from any input." Additionally, the Gemini 3.5 Flash model powers many of the new experiences, including the always-on agent Gemini Spark. Spark runs continuously on virtual machines in Google Cloud, connecting to Workspace apps and third-party services via the Model Context Protocol (MCP). It can write emails, create study guides, monitor credit card statements, and more. The Gemini app itself received a redesign called "Neural Expressive," featuring fluid animations and a new gradient interface.

Google Search Gets Its Biggest Overhaul Ever

Search is evolving into a conversational, AI-first experience. The traditional search bar now expands dynamically for longer queries and offers AI-powered suggestions that go beyond autocomplete. AI Overviews, the generated summaries, are becoming more reliable and will appear for natural language questions. Users can seamlessly transition into AI Mode, a chatbot-like interface where they can attach documents, photos, videos, and even Chrome tabs. The entire system is powered by Gemini 3.5 Flash, marking a fundamental shift in how Google handles queries.

Android XR and Wear OS: New Hardware and Software

Google provided updates on its extended reality efforts. Project Aura, a pair of smart glasses developed with Xreal, now has a redesigned compute puck with a fingerprint scanner and a lanyard for neck wear. The glasses are still in prototype stage but are expected to launch later this year with models from Warby Parker and Gentle Monster. On the wrist, Wear OS 7 introduces iPhone-style Live Updates for deliveries and sports scores, along with "Wear Widgets" that align with Android's widget formats. Select watches will also gain Gemini Intelligence, a personalized AI assistant that learns from user habits.

Gmail and Workspace: Voice Interaction and AI Imaging

Gmail Live brings voice interaction to the inbox. Users can tap an icon in the search bar and ask questions about their emails, such as finding event details or trip information. Google also launched Pics, a new AI image generation app for Workspace that allows users to click on elements of an image and leave notes for edits, streamlining iteration. Powered by Gemini and Nano Banana 2, Pics aims to reduce the friction of prompt-based editing.

AI Safety, Security, and Detection

Google is doubling down on AI provenance. SynthID watermarks and C2PA content credentials are being integrated into Chrome and Search. Users will soon be able to right-click or circle an image and ask "Is this made with AI?" This makes deepfake detection more accessible. On the security front, Google is opening up CodeMender, an AI agent for code security, to external testers. This positions Google to compete with Anthropic's recently revealed Mythos model, as both companies aim to secure high-stakes code bases. Meanwhile, OpenAI announced it will also adopt Google's SynthID for a multi-layered approach to content labeling.

AI for Science and Autonomous Agents

Google DeepMind CEO Demis Hassabis introduced Gemini for Science, a set of tools aimed at accelerating research, with the goal of "solving all disease." He also mused that humanity might be at the "foothills of the singularity." In the automotive world, Gemini will integrate with Volvo's EX60 SUV cameras to interpret parking signs and provide location-aware assistance. Google also showcased a new AI world model that can generate interactive experiences based on real-world Street View data, accessible through Project Genie.

Developer Updates and Pricing

Google AI Studio now lets users build native Android apps through vibe coding, with an embedded emulator for testing. The company also adjusted pricing for its high-end AI subscriptions: the Google AI Ultra plan now starts at $100 per month (down from $249.99), with a $200 option that includes access to Project Genie. This mirrors OpenAI's tiered approach. Additionally, Google launched a Universal Cart for AI-driven shopping that works across retailers, allowing users to add products from Search, Gemini, and eventually YouTube and Gmail, with price tracking and discount alerts.

The event also included an exclusive look inside Google's Beam Lab, where a life-size AI agent named Sophie can see and speak with visitors. While no journalists had been inside before this year, it signals Google's commitment to developing more human-like AI interfaces. As the conference wrapped, the overarching theme was clear: Google wants to be an invisible, proactive assistant that handles everything from searching the web to managing finances and even generating creative content.

Source: The Verge News