Google Search adds AI agents and Gemini 3.5 Flash model

Google announced at its I/O 2026 keynote that Search will now run “information agents” that continuously monitor the web for a user‑defined query and push updates, while the default model switches to Gemini 3.5 Flash, its newest agent‑optimized LLM [1]. The upgrade is meant to turn Search from a static answer tool into a proactive, app‑building platform, a shift that could redefine how brands appear in AI‑generated results.

New search box and agentic capabilities

The most visible change is a redesigned search box that expands as users type and accepts text, images, files, videos and Chrome tabs as inputs [1]. Beyond autocomplete, the box offers AI‑driven suggestions powered by Gemini 3.5 Flash, a model tuned for “agentic tasks and coding.” Google also unveiled the Antigravity platform, which lets a single prompt generate custom UI, interactive visuals or mini‑apps such as a wedding tracker or fitness dashboard [1]. The mini‑app feature will debut for AI Pro and Ultra subscribers in the United States, with the broader generative UI reaching all users later this summer.

Regulatory backdrop and market impact

The rollout arrives as UK regulators are forcing Google to add clearer attribution links to AI‑generated search results and to let publishers opt out of AI training data [2]. The Competition and Markets Authority (CMA) gave Google nine months to comply, but expects key controls to appear much sooner. This regulatory pressure underscores the growing scrutiny of AI‑driven search, where “AI Overviews” have been criticized for presenting confident answers without transparent source links [2]. For brands, the shift means that visibility will increasingly depend on how well their content is indexed by these agents, echoing industry warnings that “if a brand is absent in AI search, it is invisible to modern buyers” [3].

Competitive and strategic implications

Google’s move positions it ahead of competitors that still rely on traditional “blue‑link” SEO. By embedding agents that continuously track queries, Google can surface brand‑specific answers without a user re‑issuing the search, potentially capturing a larger share of the estimated 75 % of B2C and B2B decisions driven by AI search discovery [3]. However, the CMA’s requirement that opted‑out publishers cannot be down‑ranked in regular results may limit how aggressively Google can prioritize its own agents over third‑party content.

What to watch

• July 2026 – Launch of information agents for AI Pro and Ultra subscribers in the U.S.
• Q4 2026 earnings – Expectation of revenue impact from premium AI subscriptions and potential advertising shifts.
• CMA compliance timeline – Monitoring how quickly Google implements clearer attribution links and publisher opt‑out tools.

Google’s AI‑enhanced Search aims to turn queries into ongoing, app‑driven experiences, blurring the line between search and productivity tools. Whether the new agents become a decisive advantage will hinge on regulatory outcomes and how quickly brands adapt their content for AI visibility.