Google used its I/O 2026 conference to argue that the future of AI is less about chatbots and more about fast, background “agents” that quietly do work for you—but the rollout left some developers excited and others underwhelmed.

Timeline: From model hype to agent reality

Heading into I/O, expectations centered on Google’s next flagship model. Instead, CEO Sundar Pichai announced that Gemini 3.5 Pro “isn’t coming out yet,” telling a groaning audience to “give us until next month,” even as he insisted the model is already “showing great improvements.”¹

In its place, Google pushed Gemini 3.5 Flash, a smaller, cheaper model tuned for speed and autonomy. TechCrunch described it as Google’s “strongest yet for coding and autonomous AI agents,” capable of independently executing coding pipelines and even building an operating system from scratch in internal tests.² Ars Technica noted that Flash aims to make “complex agentic tasks worth doing at scale,” with performance comparable to larger models but far higher throughput.³

Pichai later promoted the launch on X, saying Gemini 3.5 Flash is “available today” across Google products and APIs and is better than Gemini 3.1 Pro “across almost all benchmarks with huge progress in coding.”Sundar Pichai (sundarpichai)⁴ DeepMind CEO Demis Hassabis called Flash “amazing,” claiming it’s 4x faster than other frontier models and “often at less than half the cost.”Demis Hassabis (demishassabis)⁵

Spark: Google’s first ‘always‑on’ personal agent

Alongside Flash, Google introduced Gemini Spark, a 24/7 personal AI agent embedded in the Gemini app. Business Insider reported that Spark can “plan parties, collate notes, and run a variety of other tasks in the background,” pitched as a shift from chatbot to proactive helper.⁶ Another report likened it to “tossing things over your shoulder” for the AI to catch and complete, from drafting status emails using Gmail and Docs to managing RSVPs and homeowner association rules for events.⁶

Google’s own marketing echoes this framing, describing Spark as a “24/7 personal AI agent that helps you navigate your digital life, taking action on your behalf, and under your direction,” built on Gemini 3.5 and designed for long‑running background tasks.Addy Osmani (addyosmani)⁷

Gemini app and agent ecosystem

To house these capabilities, Google is rebuilding Gemini into an AI hub. TechCrunch reported new features including a “Daily Brief” that pulls from a user’s inbox, calendar, and tasks to create a prioritized morning digest, plus access to a new video model called Gemini Omni and the Spark agent.⁸ The Verge highlighted a fresh “Neural Expressive” design with fluid animations and “tailored responses” that include images, timelines, and videos, and noted that Gemini 3.5 Flash is becoming the default model in the app and AI Mode.⁹

Developers inside Google are bullish: engineering leader Addy Osmani promoted Flash as “fast, great for building rich UIs + agents” and strong on “long-horizon tasks and multi-step workflows.”Addy Osmani (addyosmani)¹⁰

Consumer skepticism and access limits

Yet some observers question whether average users are ready for an AI agent ecosystem. One TechCrunch analysis argued that Google is “pitching an AI agent ecosystem to consumers who may not buy it,” pointing out that many agents—information trackers, Spark, and features like Android Halo—are either not yet shipped or initially limited to high‑end Gemini Pro and $100‑per‑month Ultra subscribers.¹¹

That paywall risks making agents a tool for power users rather than a mainstream shift in how people use the web. As Google bets its next AI wave on agents and not chatbots,² the real test will be whether users are comfortable letting always‑on systems read their email, watch their calendars, and work while they sleep—and whether the promised gains in speed and cost from Gemini 3.5 Flash translate into everyday value.