Google brings agentic AI to Search with new intelligent box

Google used I/O 2026 to make one point impossible to miss: Search is no longer just a list of links. The company said it was bringing advanced model capabilities into Search, adding new AI features, and rolling out a new intelligent AI-powered Search box it described as the biggest upgrade in more than 25 years. For agencies built around rankings, traffic, and last-click reporting, that is not a cosmetic update. It is a shift in how discovery, decision-making, and conversion may all happen inside the search experience itself.

The scale behind that shift is already enormous. Google said AI Mode has surpassed one billion monthly users and that queries in AI Mode have more than doubled every quarter since launch. Sundar Pichai also framed the moment as the “agentic Gemini era,” pointing to more than 3.2 quadrillion tokens processed per month across Google’s surfaces, 8.5 million developers building with its models monthly, and roughly 19 billion tokens per minute flowing through model APIs. Google said it now has 13 products with more than a billion users each, and that Search is where generative AI reaches more people than any other product in the world.

For agencies, the practical reading is clear: Search visibility is becoming more task-oriented and less page-oriented. Google said users can search with text, images, files, videos, and Chrome tabs in the new experience, and that conversations can continue from AI Overviews into AI Mode without a hard break. The average AI Mode search is triple the length of a traditional Search query, while planning-related queries have grown faster than overall AI Mode queries by 80% in the past six months. That changes the assignment for SEO teams. Content has to support longer, messier, multi-step journeys, not just capture a short query and a click.

The service model will likely have to change with it. Agencies that still sell only keyword tracking and traffic dashboards will have less to say as Google pushes richer prompts, multimodal input, and Search agents that can be created, customized, and managed directly in Search. Search Engine Land reported that Google’s information agents can continuously scan the web for updated information and help users along a task journey. That suggests clients will care more about whether their brand is legible to agents, whether their product data and editorial content can be parsed cleanly, and whether their assets can survive a search session that never becomes a traditional visit.

That is why the reporting layer may matter as much as the content layer. Google said more than one in six U.S. searches now use voice or images, and image searches are growing more than 40% month over month. Agencies will need to explain performance in terms beyond sessions and rankings, including how often a brand is surfaced in answer layers, how it is represented in AI summaries, and how often it remains present as users move from one prompt to the next. Search is becoming a place where users ask, compare, plan, and act in one place, and the agencies that adapt their strategy and measurement to that reality will be the ones still relevant when the clicks thin out.