Google Expands AI Mode Features in Search

Google is aggressively integrating its AI Mode into Search, now defaulting to expanded summaries on desktop and testing guided, table-based follow-up questions to keep users within its ecosystem [1, 4]. This shift aims to reduce reliance on external websites by providing direct answers and interactive tools, fundamentally altering how users navigate search results [1, 2].

Expanding the AI interface

Google has begun displaying fully expanded AI Overview summaries on desktop for certain queries, which now include a prominent "Ask anything" text box [1]. Previously, users had to click "Show more" to access these features, but the new design often occupies the entire browser window, pushing traditional search results further down the page [1]. Alongside this, the company is testing a table-based format for follow-up questions, which provides users with guided prompts to refine their search without manual typing [4].

These interface changes coincide with the rollout of "connected apps" in the U.S., allowing AI Mode to interact directly with third-party services like Instacart and Canva [2]. By enabling users to add items to a grocery cart or generate design templates without leaving the search results, Google is positioning AI Mode as a functional workspace rather than just an information retrieval tool [2]. While Google has not disclosed how companies can join this integration list, the move creates a new layer of visibility for businesses that sits alongside traditional search rankings [2].

Model performance and cost

Supporting these interface updates is the release of Gemini 3.8 Flash, the third iteration of Google’s budget model in just six weeks [3]. The model demonstrates improved reasoning capabilities, scoring 71% on the DeepSWE software engineering benchmark, up from 65.3% in the previous version [3]. However, this performance gain comes with a significant trade-off: the model’s iterative reasoning process increases the effective cost per task by approximately 40% compared to the 3.7 version, even though the base per-token price remains unchanged [3].

What to watch

• Developer adoption: Monitor whether teams migrate to the 3.8 Flash model or stick with the 3.7 version to avoid the 40% increase in effective task costs [3].
• Integration partners: Watch for additional categories added to the connected apps feature, which will signal which industries Google is prioritizing for direct AI-driven transactions [2].
• Pricing shift: Observe the transition in January 2027, when the standard token rate is scheduled to double, potentially impacting the viability of high-reasoning workloads [3].

The rapid pace of these updates suggests Google is prioritizing the entrenchment of its AI ecosystem over the immediate preservation of traditional click-through traffic. Whether this strategy successfully sustains user engagement or prompts a shift in how publishers approach search optimization remains the primary uncertainty for the coming months.