Google adds AI agents to Search

Google on May 19 added AI agents directly to Search, extending its push to put task-oriented AI tools inside one of its most widely used products. In a Search blog post published during the company’s I/O 2026 event, Google said users can now create, customize and manage multiple AI agents “right in Search,” rather than moving into a separate app experience. The company said the feature is live on desktop and mobile worldwide and begins with “information agents.” The move puts agent creation inside the familiar search box at a time when Google is also broadening Gemini across its products. Sundar Pichai, Google’s chief executive, said in his I/O keynote transcript that the company is in an “agentic Gemini era” and is focused on showing value inside products people already use every day. (blog.google) ### What exactly changed inside Search? Google’s May 19 Search post said users can now build and manage multiple agents in Search itself, with the company describing this as the start of the “era of Search agents.” The initial version centers on information agents, which Google said can keep context as users dig deeper into a topic and can surface more relevant links and supporting articles over time. (blog.google) The rollout is not framed as a limited concept demo. Google said the experience is already live across desktop and mobile worldwide, which makes Search one of the company’s broadest consumer surfaces for agent-style interaction. ### Why does putting agents in Search matter? Search has historically been where Google answered questions and pointed users elsewhere. (blog.google) Google’s update shifts that product one step further toward helping users organize ongoing tasks inside the same interface, rather than treating each query as a one-off search. That reading is supported by Google’s own wording that people can create, customize and manage “multiple AI agents for your many tasks.” CNET, in coverage tied to Google I/O 2026, said Google was reducing friction across search and other workflows as it tied AI more tightly into everyday products. That framing matches Google’s decision to surface agent management in Search instead of isolating it in a separate destination. (blog.google) ### What did Google say these agents do first? Google said it is “starting with information agents,” a phrase that points to research and information-gathering use cases rather than broad autonomous action from day one. In the company’s description, the agents retain context as a user explores and improve the relevance of links and supporting material. (cnet.com) Google did not, in the Search post surfaced here, present the feature as a full replacement for standalone agent platforms. Instead, it described an initial Search-native layer that helps people manage multiple task threads inside a product they already use. ### How does this fit with Google’s wider I/O push on agents? (blog.google) Pichai’s May 19 keynote transcript tied the company’s announcements to a broader effort to bring Gemini into more Google products and make AI useful in day-to-day workflows. Separate Google Cloud materials published this week also show the company building agent platforms for enterprise users, including tools to build, deploy, govern and optimize agents. (blog.google) That combination shows Google pushing agent tools at two levels at once: consumer-facing agents in Search and enterprise agent infrastructure in Google Cloud. The Search update is the consumer-facing piece announced at I/O 2026. ### What should users watch next? Google said the Search rollout starts with information agents, which signals further expansion if the feature gains traction. (blog.google) The next public reference point is likely to be follow-up product updates on Google’s Search and I/O pages, where the company published the May 19 announcement and Pichai’s keynote transcript. (blog.google)