Google Updates Shift Search to Task Completion

Google has rolled out a trio of search updates this week that signal a clear shift from simply delivering links to completing tasks for users. The changes, reported initially by Roger Montti for SEJ, prompted a closer look at what they reveal about the company’s evolving strategy and what it means for the rest of 2026.

At the heart of these updates is a move away from the traditional results page and toward a more action-oriented experience. Instead of just showing information, Google is now increasingly doing the work.

The most immediate update is individual hotel price tracking in Search, now live globally for signed-in users searching in English and Spanish. Users can set price alerts for specific properties and receive email notifications when rates change for their selected dates.

In a bigger move, Canvas trip planning in AI Mode has graduated from Labs preview to general availability in the U. S. This feature, first launched in November, lets users describe a trip and receive a fully customized itinerary with flights, hotels, and attractions that save automatically. Meanwhile, agent-powered store calling, which debuted in classic Search last November, is coming to AI Mode. This allows Google’s AI, using Gemini and Duplex, to call local stores directly to check inventory.

Product Leader Yao announced the updates on X, with more details in Google’s official blog post.

These announcements are not isolated. They fit a pattern visible in Google’s research, patents, and executive language since January. The company published the SAGE research paper in January, which trains agents on multi-step reasoning chains, laying the groundwork for complex tasks in Search.

CEO Sundar Pichai’s language has also sharpened. In an April interview, he stated plainly, “A lot of what are just information-seeking queries will be agentic in Search.” This followed a deep dive that tracked his shift from vague promises of change to specific descriptions of task completion. Montti argued earlier this month that this future is already here, citing the global rollout of agentic restaurant booking as proof.

A recently published Google patent, titled “Autonomously providing search results post-facto,” describes a system that waits for answers when none are available and delivers them later through assistant interactions. The pattern is consistent: Canvas expands, store calling enters AI Mode, and hotel price tracking becomes granular.

Microsoft is following the same playbook. Sumit Chauhan, President of Microsoft’s Office Product Group, announced that Copilot’s agentic capabilities are now generally available in Word, Excel, and PowerPoint. He wrote, “Copilot creates the most value when it performs the work… rather than just suggesting steps.” These features are now the default for Microsoft 365 Copilot and Premium subscribers.

The vocabulary around this shift is still fluid. Google uses “agentic” to describe features like calling and AI Mode. Pichai has described Google’s role as an “agent manager,” an orchestration layer overseeing various tasks rather than a direct competitor. Montti uses “task-based agentic search,” sometimes shortened to TBAS, as his shorthand. The market hasn’t settled on a single term, but the direction is clear.

For search professionals, these changes have immediate consequences. Local retailers face a new discovery surface. When an agent calls a store, Google’s AI, not a user, decides which businesses to contact. Google hasn’t disclosed how eligibility is determined. An analysis of 68 million AI crawler visits on Duda-hosted sites showed that sites connected to Yext, Google Business Profile, and review systems were crawled more often. Whether similar signals influence agent calls is unknown.

Hotels and travel businesses now face individual price monitoring and Canvas itineraries. There is no reporting to show if a hotel appeared in a Canvas plan or triggered an alert. Publishers face continued pressure from AI summarization. Index Exchange found that 69% of 1,200 publishers saw year-over-year declines in ad opportunities, with health and careers publishers dropping 40-50%. News and politics publishers saw only a 7% decline.

Vanessa Otero of Ad Fontes Media told Index Exchange that quality news sites remain a better experience for important events, suggesting this inventory will become the most valuable on the open web. Travel publishers, however, face a unique challenge: Canvas compiles itineraries without citing sources, making it impossible to know if their coverage influences trip plans.

Ecommerce retailers lack visibility into which stores get called, so they can’t optimize inventory feeds or Google Business Profile signals. Multi-platform coverage complicates strategy. Google’s agents favor structured data. Perplexity routes across 19 models. ChatGPT Atlas scrapes browser content. OpenAI’s Operator uses GUI vision. One business now has multiple discovery mechanisms, each with different technical needs.

The measurement gap, flagged in earlier coverage, has only widened. Search professionals cannot see if their business was included in a Canvas plan, if an agent called them, or if their hotel was surfaced in a price-tracking alert. No new reporting tools were shipped with these updates. Alphabet reported $63.1 billion in Google Search & Other advertising revenue for Q4 2025, but no new reporting has arrived to help businesses track their role in AI-mediated search.

Academic research is advancing faster than measurement. Two April 2026 papers, CW-GRPO and SKILL0, propose improvements for multi-turn search agents and agents that internalize skill packages. The training pipeline is evolving faster than the measurement pipeline. Google, OpenAI, Perplexity, and Anthropic would all need to provide equivalent agent-surface reporting. None has publicly committed.

Looking ahead, Pichai said 2027 would be an inflection point for agentic business processes. May brings Google I/O and Microsoft Build, where both companies are likely to expand their agentic surfaces. The most urgent thing to watch is reporting. If businesses can’t see their role in task-based search, they can’t optimize for it or argue about who should pay for it.

Two longer-running questions remain. Pay-per-click worked when users clicked links. Store calling, Canvas planning, and price tracking don’t produce clicks. No platform has described a replacement. Schema.org was designed for crawling, not for agents that need real-time inventory, booking availability, and action endpoints. Standards for agent-readable business data haven’t caught up.

What happens next depends on whether any platform builds reporting alongside capability. So far, none has described how it would. Until that changes, businesses will be optimizing for surfaces they can’t see. Next signals land at I/O and Build in three weeks.