Google's CEO says AI agents now brief him on the company around the clock

What Pichai said

Google CEO Sundar Pichai said he uses AI agents to keep track of what is happening across the company. As Yahoo Tech reported, he made the comments on Wednesday on the Sources podcast. An AI agent is software that can carry out a task on its own, instead of only answering one question at a time.

His problem is a familiar one for any leader. Pichai said that simply catching up on the previous 24 hours at Google could easily eat an hour of his day. Now, he said, agents gather that information and bring a summary to him.

He added that he can still dig deeper into any item, and that he often does. The agents are a first pass, not a replacement for looking at the details himself.

What the agents watch

Yahoo Tech's write-up names three areas. One is the progress of Google's AI model development. Another is how computing resources are being used across different product areas. The third is the conversation inside internal chat groups, which he can follow without reading every message.

On the computing side, Pichai described agents with access to usage data, meaning measurements of how much computing power each product consumes, how efficiently it is used and how fully it is occupied. They give him inputs and insights based on that data.

He summed up the effect in broad terms, saying the agents have helped everyone at Google do their jobs better. That is his own assessment. The report offers no figures on time saved or on how accurate the summaries are.

Why running all day matters

Pichai pointed to one feature in particular: the agents live in the cloud, meaning on remote servers, rather than on his own computer. In his words, they are running in the cloud, they're running 24/7. That means they do not stop when he leaves for the evening.

He also said they hold his context, so they can pick up where he left off. Agents can keep working on an ongoing task while he turns his attention to something else. For a person who splits the day across many topics, that is the real appeal: work continues in the background and he sees the result later.

Where humans stay in charge

Pichai was clear that people remain involved in consequential decisions. His example was how computing resources are allocated, which is among the weightier choices at an AI company. The agents inform him, but they do not make that call.

This line is worth noting because it describes a division of labor rather than a handover. Software collects, compares and condenses. A person weighs the trade-offs and decides.

The report leaves several questions open. It does not say which agents Pichai uses, what data they can reach, or which Google products power them. So it is best read as a signal about executive habits, not as a test of how reliable such agents are.

The compute backdrop

The same report recaps comments Pichai made a day earlier that explain why tracking computing use matters to him. He said Google had given up some potential Cloud revenue to put AI research first, even as Google Cloud revenue surged 82% to $24.8 billion. He said supplying computing power to Google DeepMind, its AI research lab, was not up for negotiation.

He also warned that meeting customer demand for computing capacity in 2027 would be incredibly difficult. Seen that way, an agent that monitors computing efficiency across products is tied to one of the company's tightest constraints.

The report also notes that co-founder Sergey Brin has taken a larger role in Gemini, Google's AI model family, as rivals Anthropic and OpenAI gained ground. Though he holds no formal executive role, he has pushed teams to move faster, and a former employee described his approach as startup-like. He has also urged work on recursive self-improvement, where AI helps researchers build better AI systems.