Google yesterday confirmed that information agents in Search are now available across all AI Mode languages and markets, but the rollout is gated to Google AI Ultra subscribers - a meaningful restriction that places the feature out of reach for the vast majority of the platform's billion-plus monthly users.

The announcement came on June 12, 2026, via a post on X from Robby Stein, who leads Search product at Google. According to Stein, the agents had just reached full global availability for Ultra subscribers and a broader expansion to more users was still planned for "this summer." The confirmation marks the first time the feature, introduced at Google I/O on May 19, 2026, has been described as operating across all the languages and markets where AI Mode is active - a footprint that covers more than 200 countries and territories.

What information agents actually do

At the core of this announcement is a shift in how search operates. Rather than waiting for a user to type a query, an information agent works in the background, continuously scanning the web on a user's behalf and sending a notification when something relevant surfaces.

According to Stein, a user can instruct AI Mode to "keep you updated on any topic," and the agent will "work around the clock on your behalf to send detailed updates and links to the web the moment new info is available." That is a notable departure from search as a pull mechanism - a model where the user initiates every interaction - toward something closer to a standing subscription to a topic, executed by software rather than a person.

The technical scope of what the agents monitor is broad. According to the I/O announcement by Elizabeth Reid, VP of Search at Google, information agents look across "everything on the web, like blogs, news sites and social posts, plus our freshest data, such as real-time info on finance, shopping and sports, to monitor for changes related to your specific question." That combination of structured real-time data - finance feeds, sports scores, shopping inventory - alongside crawled open web content makes these agents materially different from basic keyword alerts like those offered by Google itself through its older Alerts product.

Two examples given by Google illustrate the practical range. Someone searching for an apartment can describe their exact criteria in detail, and the agent will "continuously scan for you, notifying you when listings meet your needs." On the consumer side, a user interested in a particular athlete can instruct the agent to flag the moment that person announces a product collaboration. Both cases require the agent to maintain context about the original request, interpret new content as it appears, and make a judgment about whether it satisfies the standing query.

The Ultra restriction and what it means

Google AI Ultra starts at $99.99 per month. That is the tier at which the June 12 expansion applies. At the $19.99 per month Pro tier, users have higher access to agentic capabilities in Search, but according to the published pricing details, information agents - described as part of the agentic capabilities in AI Mode - are specifically listed as available to "Google AI Pro and Ultra subscribers this summer" in the I/O announcement. The June 12 update from Stein specifically names Ultra subscribers as the current live audience.

The distinction matters because it frames the expansion as a staged rollout rather than a general availability release. Google AI Mode itself has surpassed one billion monthly active users globally, with queries more than doubling every quarter since launch. The overwhelming majority of those users are on the free tier. The information agent capability rolling out today reaches a sliver of that total - at the $99.99 price point, paid subscribers represent a small fraction of the billion-user base.

That said, Google has been consistent about this sequencing. In the I/O announcement, Reid described information agents as launching "first for Google AI Pro and Ultra subscribers this summer," with no date attached for a broader rollout. The June 12 Stein post is consistent with that framing - a confirmation that the full-language global rollout has happened for Ultra, with Pro and general availability still ahead.

The road from I/O to global availability

The I/O announcement on May 19, 2026 was itself the culmination of a longer arc. Google has been building toward agentic search since at least late 2024. The information agent concept previewed in May represented the first time Google formally committed to deploying autonomous agents directly inside Search - not as a separate product, but as a mode of operation within the search interface itself.

At I/O, Google also replaced the default model in AI Mode with Gemini 3.5 Flash, described as "our newest Flash model delivering sustained frontier performance for agents and coding." That swap is significant for agents specifically because persistent background agents require a model capable of sustained, cost-efficient inference at scale. A model that runs continuously on millions of standing queries demands different efficiency characteristics than one that processes individual sessions. Google has not published technical details on how this works at scale, but the model choice - optimized for agentic workloads and coding - aligns with the stated design intent.

The May 19 search announcement covered five distinct areas: the model upgrade, a redesigned search box, information agents, expanded agentic booking, and Personal Intelligence. The June 12 announcement is specifically about the first of those areas reaching global completeness for Ultra subscribers.

Agentic booking and other agents on the horizon

Information agents are described as the starting point - "starting with information agents this summer" - in the I/O announcement. Google stated it is also expanding agentic booking capabilities to cover a "wide range of new tasks, including local experiences and services." Users can specify criteria for booking - a private karaoke room for six on a Friday night that serves food late is the example given - and Search will aggregate pricing and availability and surface direct links to complete the booking through the provider of the user's choice.

For categories like home repair, beauty, or pet care, Google said it would allow users to ask the system to call businesses on their behalf. These capabilities will roll out to "everyone in the U.S. this summer." Google has been building agentic calling capability since mid-2025, when the feature first appeared in testing. The expanded summer rollout represents a significant broadening of the scope.

A further category - agentic coding in Search - is also part of the summer roadmap, enabled by Google Antigravity and Gemini 3.5 Flash. This allows Search to build custom generative UI responses on the fly, assembling components like interactive visuals, tables, graphs, or simulations tailored to a specific question. Beyond one-off queries, Google described the ability to build custom dashboards or trackers for ongoing tasks - what the company described as "mini apps" for personal tasks such as planning a wedding or tracking a fitness routine. These custom experiences will be available first for Google AI Pro and Ultra subscribers in the U.S.

Subscription tiers, pricing, and what each unlocks

The subscription structure Google has established around these agent features is worth examining in detail because it governs which users get access and when.

Google AI Plus costs $4.99 per month and provides more access to Nano Banana in Search - the smaller, on-device model - alongside 400 GB of storage and 200 Google Flow Credits. It does not provide access to information agents.

Google AI Pro at $19.99 per month unlocks higher access to Gemini 3 Pro, Deep Search, agentic capabilities, and higher limits for Jules - Google's asynchronous coding agent - and Google Antigravity. According to the published pricing page, Pro subscribers in the U.S. will receive information agents this summer as part of the stated rollout schedule.

Google AI Ultra, starting at $99.99 per month, provides the highest level of access to all of the above, plus features like Project Genie - the ability to create interactive worlds using Google's Genie 3 world model - and up to 25,000 Google Flow Credits. The $199.99 per month tier provides 20x higher usage limits versus Pro. Ultra subscribers are the group Stein confirmed as having access to information agents across all languages and markets as of June 12.

The Ultra plan also includes YouTube Premium, 20 TB or more of cloud storage, and the highest limits across NotebookLM - which itself received significant upgrades on June 8, 2026, including a move to Gemini 3.5 and a cloud computer with more than 100 curated software skills.

Why this matters for marketers and publishers

The arrival of background agents in Search has direct implications for how information surfaces to users, and consequently for who benefits when it does. Analysts covering the I/O announcements have noted that existing SEO is optimized for a single moment in time - a user typing a query right now. Background agents change that dynamic fundamentally. An agent monitoring a topic on a user's behalf is not a user who returns to Search to ask again. The agent handles the monitoring, and the synthesized update it sends may or may not surface the same publishers that a manual search would have reached.

The question of which sources an agent draws on - and how it synthesizes them into the "intelligent, synthesized update" described in the I/O announcement - is not fully answered in Google's public materials. The May 19 announcement described agents looking across "everything on the web, like blogs, news sites and social posts, plus our freshest data." That phrasing includes structured Google data sources alongside open web content. The weight given to each, and the criteria for inclusion in a synthesized update, have not been publicly specified.

For paid search specifically, the arrival of agents as an ambient, persistent interface raises structural questions about the formats that can operate inside it. Google introduced mid-conversation sponsored placements in AI Mode earlier this month, a format that already attracted attention for the degree to which the sponsored label blends with the surrounding organic output. Whether - and how - advertising operates inside agent-generated notifications is not addressed in the current announcement.

The broader pattern is one PPC Land has tracked since the December 2025 experiments in which Google began testing a direct transition from AI Overviews into AI Mode on mobile. The trajectory has been consistent: a surface that began as a conversational layer on top of traditional search is progressively acquiring the characteristics of an ambient system - one that monitors rather than just responds.

The June 12 global expansion for Ultra subscribers is a relatively narrow deployment in terms of the number of users it currently reaches. But it establishes that the infrastructure works across all AI Mode languages and markets, which is the prerequisite for the broader rollout Stein described as coming later this summer.

Timeline

Summary

Who - Google, confirmed by Robby Stein, the product lead for Search at Google, via a post on X on June 12, 2026. The feature is available to Google AI Ultra subscribers globally.

What - Information agents in Google Search - persistent, autonomous agents that monitor the web continuously on a user's behalf and send synthesized updates with links when new relevant information surfaces - have been made available across all AI Mode languages and markets for Google AI Ultra subscribers. The agents operate 24 hours a day, 7 days a week, reasoning across open web content, news, social posts, and structured real-time data including finance, shopping, and sports feeds.

When - The global language and market expansion was confirmed on June 12, 2026. The underlying feature was announced on May 19, 2026, at Google I/O 2026, where it was described as launching for Google AI Pro and Ultra subscribers during summer 2026. A broader rollout beyond Ultra is still described as planned for later this summer.

Where - The agents are now live across all countries and languages where AI Mode is available - a footprint Google confirmed as covering more than 200 countries and territories following the October 7, 2025 expansion. Agentic booking expansions and generative UI mini apps remain initially limited to the United States.

Why - Google is repositioning Search from a reactive query tool into a persistent, background service that monitors topics on a user's behalf. The information agent launch is the first live deployment of that architecture at scale. The subscription gating reflects Google's staged rollout model, where Ultra subscribers serve as the initial cohort before features expand to Pro subscribers and, eventually, the broader user base. For marketers, publishers, and advertisers, the shift raises structural questions about how content surfaces inside agent-generated notifications and how advertising will operate inside a search interface that increasingly acts without a user actively submitting a query.