Google CEO Sundar Pichai uses AI agents to track daily company updates and internal progress. Speaking on the Sources podcast, he said, "The great thing about a cloud agent is that it can run without interruptions." These tools help him monitor projects efficiently, though he keeps human oversight for decisions.
Google CEO Sundar Pichai has described how he uses AI agents to keep track of information and day-to-day operations across the company. Speaking to Alex Heath on the Sources podcast, Pichai said the agents summarise the previous 24 hours of activity, monitor internal conversations and report on the progress of AI models and the use of computing resources. The setup lets him stay informed without spending hours on email and reports.
Pichai said keeping up with Google used to take about an hour a day. AI agents now handle that first pass, collecting updates and flagging what matters. He stressed, however, that human oversight remains essential, especially for important decisions such as how computing resources are allocated.
How AI agents help Pichai track Google's operations
The agents let him follow discussions in several internal chat groups without reading every message, Pichai said. They also update him on how models are progressing and how Google’s computing power is being used. When a topic needs a closer look, he can dig into the details.
Pichai stressed that the great thing about a cloud agent is that it can run without interruptions and carry context from one task to the next. This enables him to focus on new priorities while the agents continue collecting information.
He explained that as the agents handle one set of tasks, he can move on to the next problem. While this automation, Pichai said, AI agents are not meant to take major decisions; human participation is still required in critical areas.
Pichai expects AI agents to use browsers and new protocols
During the Sources podcast interview, Pichai also discussed how AI agents will access websites. He anticipates two directions for growth: agents that crawl the Web like people, and agents that communicate via specialised protocols optimised for automated communication.
Pichai referred to an elegant approach: using the browser to access a computer, because it reflects the everyday way people want to interact with websites and helps them stay engaged. However, he believes dedicated protocols will eventually offer greater scalability, although developing them will take longer.
He observed a rise in the Model Context Protocol (MCP), agent protocols, and payment standards. Google has also supported the open Universal Commerce Protocol (UCP), which aims to enable AI agents to perform commerce-related tasks across platforms that support it.
Google CEO says businesses must put users first
Pichai also spoke about the potential impact of AI agents on online businesses and advertising. In the discussion with Heath, he explained that the long-term strategy should focus on users' needs and allow businesses that offer value to remain in business.
He admitted that the agent-consumer business relationship would change in line with technological development. Pichai believes new market dynamics will arise, which could have a major impact in many industries.
The remarks follow developments by tech firms seeking to enable AI agents to navigate websites, make online purchases and perform other actions for consumers. According to Google's own documentation, Google-Agent is a user-initiated fetcher provided to agents running on Google's servers to navigate the Web and execute user-requested actions.
Despite the potential of AI agents to revolutionise how humans interact with online services, Pichai's comments indicate that humans, business interests, and interoperable standards will continue to play a role, albeit less dominant, as the technology grows.