Google is fully committed to agentic AI. During Google’s Cloud Next event on Wednesday, the technology giant unveiled numerous enhancements designed to assist its corporate clients in automating their business operations using AI agents. According to Google Cloud, 75% of its customers incorporate AI into their operations. Given the pervasive integration of AI in Google products such as Docs, Sheets, and Gmail, this high percentage is not surprising. Despite this, Google is intensifying its efforts by articulating its pursuit of the concept of an “agentic enterprise” in a blog post.
Agentic AI refers to the technology that powers autonomous agents or bots capable of executing tasks independently with minimal human supervision. It is revolutionizing AI applications, particularly in coding and workplace functions. Tech firms have embraced agents this year, showing confidence in tools like OpenClaw, Claude Code, and OpenAI’s Codex to deliver on AI’s potential to automate extensive task workflows. Now, Google is introducing a similar agentic transformation to its business technologies.
Google Cloud CEO Thomas Kurian informed journalists that agentic AI represents the direction where the company envisions AI technology heading in the future. The updates introduced this year are aimed at ensuring that customers possess secure AI processes linked to internal systems while enhancing the performance, scalability, and cost efficiency of agent operations.
The Gemini enterprise agent platform serves as the new covert technology through which businesses can manage all their AI agents. Staff members can create and deploy agents using the Gemini enterprise application, which includes a novel agent designer enabling task scheduling across various applications.
Additionally, Google is unveiling two new eighth-generation TPUs: the 8T and 8I. These cutting-edge chips are specifically tailored for tech enterprises engaged in intensive computing tasks like AI development.
The 8T chip aims to enhance training efficiency by offering three times more processing power than Google’s seventh-gen Ironwood chip. On the other hand, the 8I chip is optimized for inference tasks and boasts an 80% improvement in SRAM memory capacity while featuring around 11,152 chips within a single system.