Agentic Google search

When asked about the growth of agentic AI, Pichai said: “Just like the early days of the web, there are aspects about it that will expand access and grow the use cases, etc. I think those elements are there.”

But beyond the actual technology, he said: “We have to solve the business models for the varying players involved. There is the value proposition for all the players involved, and I think that’s going to be an equally important thing to create the ‘unlock’ here.”

Through its Project Mariner initiative, the company is beginning to integrate Gemini AI functionality into Google Search to deliver agentic capabilities, which uses natural language to assign AI agents to handle time-consuming tasks.

Robby Stein, vice-president of product for Google Search, explained in a recent YouTube video that such a capability would save people time as it would be able to perform multiple searches on, for instance, a ticket booking site, to determine how best to book seats for a group of people.

Another example is Deep Search, an advanced research tool in Google Search, which the company made available on July 16. Google said Deep Search helps researchers save time by issuing hundreds of searches, and is able to reason across disparate pieces of information to quickly produce what Google calls “a comprehensive, fully-cited report”.

Forrester principal analyst Nikhil Lai said Google appears to be copying rather than outcompeting with rivals such as OpenAI and Perplexity. “Google is spending tremendously to just keep pace with OpenAI and Perplexity, who both launched agents for deep research more than a quarter before Google,” he said.

“Now, the race to the bottom of the funnel is heating up. If Google can get a critical mass of consumers to shop on AI Mode before ChatGPT or Perplexity do, it will retain its place atop the heap of the search engine market.”