Gemini 3.5 Flash: The Engine Behind Google’s Biggest Search Revamp
Google is rearchitecting Search around Gemini 3.5 Flash, calling it the biggest overhaul of the product in roughly a quarter-century. Gemini 3.5 Flash is now the default model for Google Search AI Mode worldwide, replacing earlier Gemini releases while delivering faster responses and stronger agentic performance, particularly for coding and task automation. Users can now issue follow-up questions directly inside Search, turning one-off queries into ongoing conversations. The model supports search across modalities, so you can feed text, images, videos, files, or even active Chrome tabs into the same flow. Google is also tying Search to its Antigravity agentic development platform, allowing AI Mode to generate tailored outputs and interactive experiences instead of static summaries. With AI Mode already serving over a billion monthly users and query volume rapidly accelerating, Google is effectively repositioning Search as a front end for its most capable AI system rather than just a list of links.
From Search Box to Multimodal Command Center
The familiar search bar is turning into a multimodal command center powered by Gemini 3.5 Flash. Instead of short keywords, the field now expands to accommodate long, conversational prompts that may blend several intents at once. Crucially, it no longer limits you to typing: the multimodal search bar accepts voice, images, videos, documents and browser tabs in a single query. This flexibility lets people, for example, drop in a screenshot, attach a file, and add clarifying text to get a unified, AI-driven response. Once you submit, Search can render richer interfaces on the fly, such as interactive visuals, tables, graphs or simulations alongside traditional results. If you prefer classic web pages, a dedicated Web view still surfaces standard link lists. By merging conversational AI Mode with AI Overviews and this new interface, Google is trying to remove the mental friction of deciding which experience to use every time you have a question.
Information Agents: Always-On AI Assistants Inside Search
One of the most transformative new agentic search features is information agents, persistent AI agents that run in the background around the clock. Instead of repeatedly searching for updates, you describe your goal once—such as tracking apartment listings that match specific criteria, following certain stocks under particular conditions, or monitoring news about a favorite athlete’s sneaker collaborations—and the information agent keeps watch across the web. It scans blogs, news sites, social posts and real-time streams in finance, shopping and sports, then notifies you when something relevant changes. These AI agents search continuously, maintaining context about your preferences and constraints across sessions. Early access will arrive first for AI Pro and Ultra tiers, but the concept signals a shift from reactive queries to proactive, ongoing assistance. Search is evolving from a static answer box into a persistent digital researcher that quietly works on your behalf between visits.
Agentic Coding and Mini Apps Turn Results into Live Tools
Gemini 3.5 Flash also introduces agentic coding in Google Search, effectively letting users transform results into live mini apps. Through Antigravity, Search can auto-generate custom interfaces—dashboards, trackers or interactive widgets—tailored to specific tasks. Generative UI capabilities assemble components like charts, tables, maps and simulations in real time, based on what your query needs. If you’re repeatedly planning workouts, managing a move or organizing an event, Search can build a persistent tracker that pulls in fresh data such as reviews, local information, live maps or weather. You can return to these experiences over time, updating inputs while the underlying AI agents keep them synced with the latest information. For power users, custom coding experiences allow deeper control over how these tools look and behave. These AI agents search and compute behind the scenes, turning Search into a lightweight app platform rather than merely a destination for links.
Personal Intelligence and the Future of AI Agents in Search
To make AI agents in Search genuinely useful, Google is expanding Personal Intelligence in AI Mode to more people across many languages. With user consent, Gemini can draw on signals from services like Gmail and Google Photos to add personal context to responses, while still grounding answers in live information from the open web. Combined with information agents and agentic booking features that can call businesses for home repair, beauty or pet care appointments, Search starts to resemble a general-purpose command center for everyday tasks. Agentic booking is set to reach more users soon, and the same underlying infrastructure powers Universal Cart and other transactional features. Together, these updates mean many tasks—research, planning, monitoring, even simple app-like workflows—can be initiated, managed and completed entirely within Google Search, without switching tabs. The search bar is becoming the place where AI agents search, decide and act on your behalf.