On May 19, 2026, Google used I/O 2026 to make Search a more explicit AI agent surface. The company said Search is entering the era of Search agents, starting with 24/7 information agents inside Search, broader agentic booking for local experiences and services, business-calling features in the U.S., and custom mini apps that Search can generate on the fly.
That matters because this is not just another AI search-box update. Google is moving agents into one of the highest-intent products on the internet, where people already research purchases, compare providers, plan trips, solve problems, and decide what to do next.
What Google actually launched in Search
The headline feature is information agents. Google says users will be able to create, customize and manage multiple agents directly inside Search. These agents are designed to run in the background, reason across changing information on the web, and send synthesized updates when something important changes.
Google’s examples were deliberately practical: apartment hunting, tracking product drops, and monitoring niche topics that normally require repeated searches. The bigger product shift is that Search is being recast as an always-on monitoring and action layer rather than a place for one-off queries.
Google also said Search is expanding agentic booking to more tasks, including local experiences and services. For select categories such as home repair, beauty and pet care, Google will also let users ask Search to call businesses on their behalf in the U.S. this summer. That pushes Search closer to a workflow surface that gathers requirements, checks availability and helps finish the transaction.
A third piece is easy to miss but may matter just as much: Google is bringing Antigravity-powered generative UI and mini apps into Search. Instead of returning only links or summaries, Search will be able to assemble interactive tools, dashboards, trackers and simulations on demand. Google framed these as custom experiences that can persist across longer tasks such as planning a wedding, moving house or managing a fitness routine.
Why this is bigger than a smarter search box
Google paired the product launch with scale numbers that make the announcement harder to dismiss as a demo. The company said AI Mode has now surpassed 1 billion monthly active users globally, and that AI Mode queries have more than doubled every quarter since launch. It also said planning-related AI Mode queries have grown faster than AI Mode queries overall in the last six months.
Those details matter because they show Google is not adding agents to an empty product. It is layering agent behavior onto a search surface that already has massive usage and high commercial intent. In practice, that gives Google a distribution advantage many standalone AI agent products do not have.
The strategic shift is from answer retrieval to task continuation. An information agent can keep watching after the initial query. An agentic booking flow can keep narrowing options after the first prompt. A custom mini app can keep organizing work after the first result page. That is a different product model from classic search and a different adoption path from separate chatbot apps.
Business impact lands first in discovery, intake and local service flows
For businesses, the immediate implication is that Google is trying to sit closer to the action layer of customer demand. If Search can monitor, compare, call, summarize and route users toward next steps, then customer acquisition may increasingly depend on whether a business is easy for agents to understand and transact with.
- Local service companies may feel this first. If Google agents start collecting availability, checking fit and calling providers on a user’s behalf, then intake quality, response speed and structured service data become more important than a traditional landing-page pitch alone.
- Commerce teams should watch how agentic booking and related shopping flows change where comparison happens. Google is clearly trying to keep more planning and decision-making inside its own agent layer before handing users off.
- AI and automation teams should pay attention to the mini-app angle. If Search becomes a place where users can spin up lightweight trackers, planners and decision tools, the line between search, software and agent orchestration gets thinner.
There is also a broader platform implication for agent builders. Google is not only shipping another model. It is using Gemini 3.5 Flash and Antigravity to insert agent behavior into distribution surfaces people already trust and use every day. That is a more consequential move than a benchmark win because it can change where AI workflows begin.
What to watch next
The near-term watchpoints are rollout and control. Google said information agents will launch first for Google AI Pro and Ultra subscribers this summer. It also said business-calling capabilities for select local-service categories will roll out to everyone in the U.S. this summer, while custom Search experiences and mini apps will start with Pro and Ultra users in the U.S. in the coming months.
The bigger question is how much of the customer journey Google keeps inside its own surfaces. If Search agents become the default place where users compare options, track changes and initiate service requests, businesses may need to optimize not just for ranking, but for agent readability, agent actionability and handoff quality.
The practical takeaway is straightforward: AI agents are no longer confined to separate copilots or enterprise demos. Google is inserting them into the world’s most important discovery interface, which means search strategy, service operations and workflow automation are starting to converge.